Hindawi Publishing CorporationISRN RadiologyVolume 2013 Article ID 529463 18 pageshttpdxdoiorg1054022013529463

Review ArticleMagnetoencephalography Fundamentals and Established andEmerging Clinical Applications in Radiology

Sven Braeutigam

Oxford Centre for Human Brain Activity Department of Psychiatry University of Oxford Warneford Hospital Oxford OX3 7JX UK

Correspondence should be addressed to Sven Braeutigam svenbraeutigampsychoxacuk

Received 11 June 2013 Accepted 3 July 2013

Academic Editors H Akan W Chen S Ramcharitar and V D Souftas

Copyright copy 2013 Sven Braeutigam This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Magnetoencephalography is a noninvasive fast and patient friendly technique for recording brain activity It is increasingly avail-able and is regarded as one of the most modern imaging tools available to radiologistsThe dominant clinical use of this technologycurrently centers on two partly overlapping areas namely localizing the regions from which epileptic seizures originate andidentifying regions of normal brain function in patients preparing to undergo brain surgery As a consequence many radiologistsmay not yet be familiar with this technique This review provides an introduction to magnetoencephalography discusses relevantanalytical techniques and presents recent developments in established and emerging clinical applications such as pervasivedevelopmental disorders Although the role of magnetoencephalography in diagnosis prognosis and patient treatment is stilllimited it is argued that this technology is exquisitely capable of contributing indispensable information about brain dynamicsnot easily obtained with other modalities This it is believed will make this technology an important clinical tool for a wide rangeof disorders in the future

1 Introduction

Magnetoencephalography (MEG) is a noninvasive techniquefor recording brain activity MEG was first introduced tothe scientific community in 1972 [1] and it has undergonesubstantial technological advances ever since Modern mul-tichannel whole-head systems provide reliable fast andpatient friendly scanning that is increasingly being used forclinically oriented research into a wealth of mental disordersand abnormal conditions such as adult and pediatric epilepsy[2ndash6] autism [7 8] schizophrenia [9] Williams syndrome[10] Landau-Kleffner syndrome [11] Alzheimerrsquos disease [1213] depression [14] attention deficit hyperactivity disorder[15 16] and dyslexia [17] Moreover MEG has been used tostudy neuronal change and reorganization following stroke[18] head trauma [19] and drug administration [20] MEGresearch centers now exist in many countries with perhapsJapan USA Germany UK and Finland leading in terms oftotal installations MEG has been approved for clinical evalu-ation by FDA (Food andDrugAdministration) andMedicarein the USA where many insurance companies presently arecovering MEG scans in patients with epilepsy intracranial

neoplasia and vascular malformations [6] Typically MEGscans have to be coordinated on a case-by-case basis requiringefforts on the part of the MEG center the patientrsquos doctorsand the patient Centers elsewhere are beginning to acceptMEG scanning request from doctors and other healthcareprofessionals although practice and support may vary con-siderably between countries

Despite increasing availability of high-end scanners how-ever the dominant clinical use of MEG currently centers ontwo partly overlapping areas namely localizing the regionsfrom which epileptic seizures originate and identifyingregions of normal brain function in patients preparingto undergo brain surgery As a consequence many radiolo-gists may not yet be familiar with this powerful techniqueThe purpose of this review is to provide the reader withan introduction toMEG fundamentals and relevant analyticaltechniques including an exposition of modern approaches toartifact correction an issue of particular importance to clin-ical studies Although extensive this paper does not attemptto discuss every possible clinical application ofMEG Insteademphasis is given to recent developments in two broad areasnamely epilepsy and autism The former area is rather well

2 ISRN Radiology

established implying a reasonably defined role of MEG indiagnosis treatment and patient management whereas thelatter is an emerging field forMEG reflecting a growing trendin using neuroimaging in the study of developmental disor-ders

2 MEG Technology

MEG is based on the detection of the magnetic fields that aregenerated by the currents flowing in neurons Unlike positronemission tomography (PET) or functional magnetic reso-nance imaging (fMRI) MEG does not rely on secondaryeffects induced by brain activity but directly measures themagnetic fields primarily generated by postsynaptic neuronalionic currents In contrast to its closest cousin the elec-troencephalogram (EEG) MEG signals are reference-freeand essentially unaffected by conductivity differences on themagnetic flux providing an almost undistorted view of brainactivity and simplifying data analysis and interpretation ofpatterns MEG is mainly sensitive to magnetic fields gener-ated in the cerebral cortex but modern whole-head systemswith multiple sensor configurations can detect activity insubcortical regions Typically neural magnetic field changesare extremely small ranging from 10minus14 T or even lessfor evoked fields to approximately 10minus12 T during interictalepileptic spikes [21 22] By comparison a typical fMRI scan-ner operates at 15 T and the Earthrsquos magnetic field is approx-imately 10minus4 T Currently the only practical and competitivedetectors capable of recording such small field changes aresuperconducting loops that are coupled into superconductingquantum interference devices (SQUID) that respond to thechanges inmagnetic flux through the loops [23 24] AnMEGscanner is illustrated in Figure 1

The loop configuration determines what the detectormeasures for example a simple planar coil measures thechange in the magnetic field perpendicular to the loop planeand a figure-of-eight-shaped coil measures the change in themagnetic field gradient An analogy is in orderThe loop (coil)can be compared to a lens forming an image dependent on thefocal length where the SQUID transforms themagnetic fieldsinto electrical voltages much like a camera sensor transformsan optical image A typical MEG imaging system has severalhundred sensors surrounding a head shaped recess in a liquidhelium-filled Dewar (cryogenic temperatures are requiredfor SQUID operation) The whole instrument is housed ina magnetically screened room that is required to reduce theambient magnetic field noise The best instruments currentlyavailable achieve noise levels approaching 10minus16 T per squareroot of the bandwidth over which the signals are recordedThe temporal resolution of MEG is less than a millisecondwhich is sufficient to trace even the fastest neuronal processesThe spatial resolution of MEG depends on signal quality andcan be as low as a few millimeters for appropriate sourcemodels which is in the range of subdural electrode strips andgrids [6 25]

There are subtleties associated with the sources that aredetected Even with modern scanners the fields emanatingfrom single nerve cells are too weak to be measured outsidethe head Similar to EEG this implies that detectable field

changes require large numbers (few thousand to a few tenthousands) of neighbouring cells to carry aligned currentsof suitable strength and duration to allow for spatial andtemporal and spatial summations It is commonly assumedthat these conditions are met through postsynaptic currentsrather than currents associatedwith action potentials or otherelectrobiochemical processes [26ndash28] and [29 Chapter 1]MEG is technically challenging however this technologyhas some significant advantages It is a passive and con-tactless measurement that is completely non-invasive easyto prepare and minimally demanding on the subject Theinstrument is silent in its operation where the MEG detectorrecords the brain activity continuously as the subject per-forms tasks required by the investigator Longitudinal studiesare ethically possible MEG is capable of providing meaning-ful results on its own although information extracted fromother brain imaging techniques and coregistered with theMEG can be used to guide analysis and interpretation

3 Analytical Approaches

The primary outputs provided by any magnetoencephalo-graphic scanner are readings of the variation in time of themagnetic fields produced by electrical activity in the brainand detected over the head These signals are analyzed invarious ways correlated with experimental conditions andinterpreted within the framework of a given clinical inves-tigation where often findings obtained from other imagingmodalities are considered or even integrated into the analysisA plethora of analytical techniques to extract informationfrom the MEG signals have been developed and new meth-ods continue to emerge rapidly Currently the abundance ofapproaches implies an absence of standard protocols forMEGanalysis but there is a clear drive towards standardizationand guideline papers are beginning to emerge [30] Despitethe variety and partial incomparability of methodologieshowever certain approaches have crystallized that are usedfairly consistently across clinical studies

31 Event-Related Fields Event-related fields (ERF) are directanalogues of the perhaps more familiar event-related poten-tials (ERP) Event-related approaches rely on repeatedlypresented auditory visual tactile electrical or other stimuliThe appearance of each stimulus marks a point in timeusually denoted as stimulus onset with respect to which anepoch of data is defined The signal observed in the post-stimulus interval is assumed to be independent of stimulusrepetition but to be contaminated by random brain signalsconsidered to be noise This assumption is often justified butmay have limited validity in studies of higher order functionWithin the assumption of independence against repetitionthe stimulus-specific neural response can be recovered fromthe measured signal by averaging the data with respect tostimulus onset The resultant waveform is usually called anevoked response The process of averaging may be extendedover several subjects to yield grand mean waveforms Typi-cally the stimulus and subject dependence of peak amplitudesand latencies of the evoked responses allow some insight intotheir functional significance [31]

ISRN Radiology 3

Elektra NeuromagTM

3 seconds

600 f

Tcm

Figure 1 The MEG scanner at the Oxford Centre for Human Brain Activity Left Shown is a typical set up where the subject is seatedupright under the scanner A button box is used to record behavioral responses and an eye-tracker supplies additional psychometrical dataMeasurements in supine position are possible and a doctor or nurse can stay within the magnetically shielded room in order to monitor thepatient if needed Middle The brain magnetic fields are recorded with a helmet-shaped array of 102 SQUID devices featuring 3 pick-up coilseach (306 channels in total) Two first-order gradiometers measure two orthogonal spatial gradients of the magnetic field in longitudinal andlatitudinal directions respectively These channels are most sensitive to the tangential neuronal currents in the region below the device Amagnetometer coil (black rectangle) measures the magnetic field and is sensitive to deeper sources Right Typically the cortex-to-detectordistance is 4-5 cm The primary output of the scanner is a trace for each channel representing the magnetic field or gradient as a function oftime (for presentation only 16 channels are shown brain image courtesy of Elekta Neuromag OY Helsinki)

32 Time-Frequency Analysis Viewing brain activity as anoscillatory phenomena there is a long history of studyingfundamental brain rhythms such as theta (4ndash7Hz) or alpha(8ndash13Hz) waves that under certain conditions may indicategeneral states of the brain for example increased alpha isoften seen when subjects close their eyes [32] Robust algo-rithms are now available to analyze neurophysiological sig-nals in terms of oscillations that vary in frequency and timeTypically transient oscillatory brain activity is consideredin the context of well-defined stimuli and is identified aseither an evoked oscillatory response appearing at the samelatency and phase in each trial or an induced oscillatoryresponse appearing with a jitter in phase from one trial toanother and therefore not detectable in the averaged dataBoth types of response are assumed to provide informationabout brain dynamics beyond what can be extracted from theconventional evoked response which includes all frequencies[33 34]

Oscillations at higher frequencies in the beta (15ndash30Hz)and gamma (gt30Hz) bands have received much attention inrecent years Experimental data reveal that synchronizationof activity in these bands is involved in processes such asworking memory semantic processing and sensory-motorintegration and that synchronization may be altered underpathological conditions [35 36] Theoretical considerationssuggest that synchrony of oscillations can facilitate thecoordinated interactions of large neuronal population dis-tributedwithin and across distinct regions of the brain where

interneuron networks are believed to play an important role[37 38]

Of particular relevance to clinical application is event-related desynchronization (ERD) ERD refers to the well-known phenomenon that certain events or stimuli can desyn-chronize or even block parts of the ongoing brain activityThese induced changes can be detected using a techniqueemploying band-pass filtering squaring and averaging inorder to obtain a measure of the oscillatory power in a givenfrequency band of interest ERD is then quantified relativeto the oscillatory power obtained from a reference intervalThe ERD measure may become positive dependent on thenature of task and stimuli in which case one uses the termevent-related synchronization (ERS) ERD (ERS) may beconsidered to be due to a decrease (increase) in synchronyof the underlying neuronal populations and is recognized asa robust marker of normal and abnormal neuronal processesin clinical applications of EEG and MEG [39]

33 Source Estimation Most relevant to clinical MEG issource estimation that is the identification of the brain areasgenerating a given signal recorded by the detectors (alsoknown as magnetic source imaging MSI) Source estimationis ultimately based on the theory of electromagnetism whichis governed by 4 equations known as Maxwellrsquos equationswhich relate the spatial and timederivatives of the electric andmagnetic fields to the charges and currents present within agiven volume defined by its geometry electric permittivity

4 ISRN Radiology

and magnet permeability [40] In the case of electrophysi-ological signals the useful frequency spectrum is typicallybelow 2 kHz and most studies deal with frequency between005Hz and a few hundred hertz As a consequence onecan disregard the time derivatives yielding the quasi-staticapproximation

119861 (119903) sim int119871 (119903 1199031015840

) times 119869 (1199031015840

) 1198891199031015840

(1)

also known as Biot-Savartrsquos law The rule relates the currentflow 119869 at location 1199031015840 within a given source volume to the mag-netic field 119861 observed at position 119903 The integral (continuoussum) runs over the volume containing the currents where thelead field 119871 establishes a link between currents and magneticfield taking into account volume geometry and propertiesTypically one associates the magnetic field with the actualmeasurement in which case the lead field is also dependenton the type position and orientation of the detector usedThen the physical interpretation of the lead field is that itrepresents the sensitivity pattern of a given detector and agiven source volume [21 41]

Conceptually the law constitutes a forward problem thatis knowing the currents and the properties of the embeddingmedium one can calculate the magnetic field in an analyticalor numerical fashion In contrast every source estimationprocedure has to face the so-called inverse problem that isthemathematical transformation that uses the observedmag-netic field as input and determines the sources of the signalsHere there is a problem that is shared with many algo-rithmic proceduresmdashthe inverse problem in MEG is deeplymathematically ambiguous [41 42] No matter how highquality the data is and howmany measurement channels oneuses there can always be silent albeit neurophysiologicallyrelevant currents in the brain that are not recorded by thedetectors The MEG inverse problem is approached prag-matically by simplifying the description of the sources orby adding additional physiologically reasonable assumptionsThe forward model is always needed for source estimationand it is commonly assumed that the current flow in thehead volume can be expressed as the sum of a primary orimpressed current due to the electromotive forces associatedwith biologicalmainly synaptic activity in neural tissue and apassive current flow that results from motion of extracellularcharge carriers [43] Consider

119869 (119903) = 119869impressed (119903) + 119869passive (119903) (2)

331 Equivalent Current Dipoles The arguably simplest anduntil recently most commonly used source estimation tech-nique is the use of equivalent current dipoles (ECD) An ECDcan be viewed as the spatial average of all impressed currentswithin an area of the brain assuming that the arearsquos spatialextent is small compared with the distances to the detectorsEach equivalent dipole is characterized by 6 parametersdescribing the dipolersquos position orientation and strengthwithin the head In general a nonlinear iterative algorithmis applied in order to find the parameters best describingthe data measured In each iteration step the magnetic fields

generated by the dipoles are compared with the fields mea-sured by the detectors and the parameters are adjusted tominimize the field difference quantified by a cost functionTypically a single homogenous sphere is used to approximatethe head volume although other conductormodels have beenused such as ellipsoids several spheres each fitted to thelocal curvature of the head or realistically shaped geometriesbased on individual structural scans

In case of the homogenous sphere the magnetic forwardproblem becomes independent of tissue conductivities andthe spherical symmetries imply that radially oriented currentdipoles become hidden Thus this source model can onlyaccount for tangential dipoles assumed to be primarilylocated in the walls of cortical fissures and sulci but not in theconvexial cortex [21 41 44] ECD approaches vary accordingto the number of dipoles employed (one dipole being themost common choice) the number and type of parametersallowed to change during fit (eg fixed position with rotatingorientation and time-varying strength) and the nature ofsearching or scanning algorithm used to fit parameters todata

332 Imaging Methods Imaging approaches to the inverseproblem explore quasi-continuous estimates of activation bydividing the brain volume into distinct voxels and allocating3 orthogonal equivalent current dipoles to each voxel Severalthousand voxels are usually employed approximating anyarbitrary spatial distribution of post-synaptic currents in thebrain The inverse problem in this case is linear since thesource locations are known leaving only source amplitudes tobe estimated The problem however is severely underdeter-mined given that the number of detectors is small comparedto the number of voxels and regularization is required inorder to restrict the range of allowable source currents whilesimultaneously minimizing the difference betweenmeasuredand predicted magnetic fields

Symbolically

minimum119869

119861 minus 119871 times 119869119901 + 1205821003817100381710038171003817119882model times 119869

1003817100381710038171003817

119901

(3)

where sdot sdot sdot 119901 denotes a norm (roughly strength) of thequantity enclosed In other words one estimates for all voxelsthose current amplitudes that best explain the data and keepa chosen model term also known as a penalty term smallCommonly the voxels are not uniform throughout the cra-nium but restricted to the grey matter or more specifically tothe cerebral cortex In line with the assumption that currentsin the dendritic trunks of pyramidal cells predominantly con-tribute to external fields the dipole orientation may furtherbe constrained to be perpendicular to the local cortical sur-face The parameter 120582 controls the amount of regularizationand its optimal value is usually determined in a data-drivenfashion using an appropriate heuristic approach The weightmatrix 119882model is used to assign additional properties of theinverse solution such as focal sources In essence the weightand the normparameter119901 together define the type of imagingsolution each of which has its merits and disadvantages [44ndash50] The most common choices of norms and weights aresummarized in Table 1

ISRN Radiology 5

Table 1 Comparison of distributed source algorithms often used in clinical MEG research

Name 119901 119882model Comment

MNE 2 1+ simpleminus currents packed towards surface of volume

MNE (weighted) 2 1current-gain + no surface biasminus blurry images

LORETA 2 Laplacian+ + deep source accurateminus estimates superficial sources too deep

MCE 1 1current-gain + focal estimatesminus strong sensitivity to noise

MNE minimum norm estimate (eg [42]) MCE minimum current estimate [46] LORETA low resolution brain electromagnetic tomography [45] +Math-ematical operator representing flux density and gradient flow of a function

333 Beamforming Beamforming approaches to the inverseproblem aim at discriminating between signals originatingfrom a location of interest and those generated elsewhere inthe brain Similar to imaging techniques one assumes thatthe MEG measurement can be modeled by a set of fixedcurrent dipoles distributed throughout the brain In contrastto imaging techniques however one does not attempt to solvea set of underdetermined equations using norm constraintsInstead a beamformer estimates the neural activation at eachvoxel independent of all other locations by means of a spatialfilter Ideally such filter 119865 has complete spatial specificityzeroing the lead field at all locations except the chosen one

119865 (1199030

) times 119871 (119903 1199031015840

) =

1 1199031015840

= 1199030

0 1199031015840

= 1199030

(4)

where 1199030

and 1199031015840 are locations with the source volume and119903 refers to the location of measurement Once the filter hasbeen estimated the neuronal current at position 119903

0

can berecovered from the magnetic fields detected by the sensorarray

119869 (1199030

) = 119865 (1199030

) times 119861 (5)

This expression is commonly interpreted as a virtual electrodethat is one gains knowledge about a current as if an electrodewas attached to the location of interest In practice it isgenerally impossible to have complete attenuation in the stop-band that is locations other than the one of interest and oneaims for designing a filter that is optimal in some sense[44 51]

Arguably the most influential guiding philosophy fordesigning an optimal filter is the linear constrainedminimumvariance approach or LCMV for short [51] The idea is todetermine a filter that minimizes the variance of the sourceat a given location while satisfying the linear response con-straint 119865 times 119871 = 1 ensuring that the signals of interest passby the filter Minimization of source variance which is ameasure of current strength optimally allocates the stop-band response of the filter tominimize the contribution to theoutput due to signals generated by strong sources in the stop-band The mathematical apparatus of beamforming relies onthe assumption that the dipole sources generate mutuallyuncorrelated brain activity Correlated activation cannot bereliably disambiguated from correlated signals at the sensor

level due to volume conduction and field propagation Inother words the sensor readings might be correlated becauseunderlying activations are correlated or the field generated bya source is detected bymore than one sensor at the same timeThe assumption that brain activity is uncorrelated acrossregions might not be physiologically justifiable however ithas been shown that beamformer source estimates are tosome extent robust against correlated sources [44 51 52]

A related approach is known as hardware beamformingexploiting the physical properties of certain detectors typesSensors like planar gradiometers employ two coils measuringorthogonalmagnetic field gradients (spatial variation) at eachsensor location The root-mean-square of the two readingscan be interpreted as the strength of a current dipole locateddirectly below each gradiometer yielding a sensor-level mapof brain activation that can be projected onto the corticalsurface The method is easy to implement and might beapplicable even if the scanner does not have sensors directlysuitable for hardware beamforming [53]

334 Summary Source Estimation The researchers and clin-icians can now draw on awide range of source reconstructionmethods often available in form of commercial and opensoftware packagesThe source images obtained fromdifferentalgorithms are often broadly consistent and may in favor-able circumstances offer meaningful insights into functionalanatomy at the centimeter level However no general sourceimaging method is currently available where the results maydepend on the data and method of analysis In particular thespatial extent of source images may reflect the properties ofthe estimate rather than the true extent of the neural currentStatistical inferences based on source estimates need to bejudged with care as significance does not necessarily implysignificantly nonzero real primary current in a voxel [54] Inany case the interpretation of source estimates should includeconsideration of the nature of the underlying algorithm

4 Artifacts and Artifact Removal

The high sensitivity of MEG implies that the recordings aresensitive to artifacts originating from a variety of physiolog-ical and environmental sources This issue is not trivial andcan become a serious challenge in certain patient populationsdue to amongst others heightened physiological responses

6 ISRN Radiology

lack of control over movements and magnetic materials nec-essary for diagnosis and treatment [4] In general the impactof artifacts may be reduced or controlled to some extentthrough careful preparation and experimental design andexclusion of the most affected data may facilitate analysis andinterpretation However it is impossible to exploit clinicaldata fully without advanced approaches to signal processingthat can reliably separate brain signals and artifacts (Figure 2illustrates the concepts discussed in this section)

41 Physiological Artifacts Arguably the two most dominantartifact sources are the heartbeat and eye blink Althoughthese physiological artifacts can have rather large amplitudessignal separation is challenging as the artificial signals mayresemble activity originating from neuronal sources Regard-ing the eye normal automatic blinking occurs between about5 and 20 times per minute and causes magnetic fields overpredominately frontotemporal regions that can be muchlarger than the magnetic fields of spontaneous and evokedbrain activity The primary artifact is generated by currentflow between the eye and scalp and by excitation of ocularmuscles controlling eye and eyelid movements The artifacttime course recorded by MEG or EEG during a spontaneousblink corresponds to the eyelid motion reaching a closedposition on average 120ms after the blink is initiated andbefore returning to a fully open position approximately240ms after closure In addition to the direct artifact fieldsblink related neuronal activity lasting up to several tenths ofa second is also often observed over posterior parietal andfrontal cortices as well as other areas of the brain associatedwith higher-order processing [57 58] For artifact correctionone generally obtains an independent record of the artifactby measuring the electrooculogram (EOG) with electrodesplaced close to the eyes

Regarding the heart electrical currents moving throughthe organ during a heartbeat give rise to magnetic fields thatare over the chest between two and three orders of magni-tude larger than the brain magnetic fields The time course ofthe cardiac artifact closely corresponds to the propagation ofelectrical activity through the heart muscle Each heartbeatbegins with an impulse from the sinoatrial node the heartrsquosmain pacemaker activating the upper chambers (atria) Thisactivation is seen as the P wave (duration about 80ms) in theelectrocardiogram (ECG) Next the electrical current flowsdown the heart and activates the lower chambers (ventricles)giving rise to a characteristicwave known as theQRS complex(80ndash120ms) Subsequently the electrical current spreadsback over the ventricles in the opposite direction generatinga recovery wave commonly denoted by the letter T (160ms)Typically these currents flowing in the heart are observableto varying degrees in MEG and EEG recordings causingartifacts that may resemble epileptic spikes and slow waves Itis generally assumed that secondary artifacts due to heart beatrelated blood pulsation body movements (ballistocardio-effects) and susceptibility changes are negligible but mayexist in MEG and EEG recordings if the lungs skin or bloodvessels contain magnetic particles [59 page 162] and [60]For artifact correction one generally obtains an independent

record of the artifact by measuring the ECG with electrodesplaced on both arms close to the wrists

Various approaches have been suggested in the past tocorrect for these artifacts often but not exclusively utilizingequivalent current dipoles to model nonneuronal aspectsof the artifacts for example the retinarsquos electrical charge[61] Despite merits however those methods are increas-ingly being replaced by signal projection-based approachesProjection methods are broadly analogous to obtaining a2-dimensional shadow from a 3-dimensional object wherethe direction of illumination determines which features arevisible in the projection Mathematically the data are repre-sented as points (also called vectors) in a space of dimension119899 equal to the number of measurement channels Each 119899-dimensional vector corresponds to one sampled time pointThen the set of new vectors 119881 is calculated to form a basisfor the unwanted signal features that is the artifact to beremoved In other words the set 119881 spans a subspace that isas far as possible occupied by the artifact but not the neuro-physiological signals of interest An artifact-free (clean)measurement is readily obtained through projection

119872119888

= (1 minus 119881 times 119881+

) times 119872 (6)

where 119872 denotes the measurement matrix of dimension119899 times (number of time-points) 1 denotes the 119899-dimensionalidentity matrix and 119881+ is the pseudoinverse of 119881 which hasdimension 119899 times 119896 where 119896 is less than 119873 [62] Dependenton the artifact to be removed 119896 typically assumes a valuebetween 1 and 3

Perhaps the most powerful method to determine theartifact subspace 119881 is a type of blind source separation(BSS) known as independent component analysis (ICA) ICAis entirely data driven and assumes that the observations119872 constitute a linear superposition (mixing) of statisticallyindependent source signals The number of independentsources 119904 is assumed to be the most equal to the number ofmeasurement channels A frequently quoted analogy is theldquococktail party problemrdquo In this scenario party guests holdmultiple simultaneous conversations recorded using micro-phones located at fixed positions around the room Knowingthe number of speakers (s le number of microphones) ICAallows recovering of the individual conversations by unmix-ing the recordings

119878 = 119880 times119872 (7)

where the number of rows of 119878 is equal to the number ofspeakers Each row of 119878 contains the sound from one speakeronly The number of columns of the un-mixing matrix 119880 isequal to the number of microphones and each element of119880 can be interpreted as a weight determining how much anindividual speakerrsquos voice contributed to a given microphone[63 64] Using this analogy the task is then to identify theldquospeakersrdquo who deliver the artifact and guide them out theroom

Over the last 7 years or so several algorithms have beenpublished in order to address this issue in an automatedfashion The main feature that distinguishes the differentapproaches is whether or not an independent EOG andor

ISRN Radiology 7

EOG

CorrectedOriginal

Source power

Original Corrected

Front

Back

RightLeftSi

gnal

(fT

cm) 300

200100

0minus100

minus200

Time (s)4 5 6 7 8

(a)

(b)

(c)

Figure 2 Artifact removal (a) Shown are data from one channel before (black) and after (red) ICA-based correction for eye-blink artifactsidentified by EOG activity (bottom trace) The inset illustrates the 2-dimensional artifact space used for projection (time-courses and spatialtopographies) The first dimension is dominant Note the artifact has similar features as epileptic discharges in that it involves several MEGchannels has sharp peaks and stands out from ongoing background activity (see also [55]) (b) Equivalent current dipole modeling of anevoked somatosensory response in a child before (left) and after (right) SSS-based movement correction After correction the ECD assumesa physiologically plausible location (courtesy of Elekta Neuromag OY Helsinki) (c) Brain activity can be localized accurately despite strongartifacts caused by DBS electrodes (right) Before beamforming-based correction strong nonphysiological interferences outside the brainare observed (left adapted from [56])

ECG recording is needed for artifact identification Table 2shows a brief overview of some of the relevant studies

A recent study in 26 healthy volunteers compared variousartifact detection metrics as well as different flavors of ICAand other blind source separation techniques [70] In gen-eral the authors found automated artifact removal methodspracticable where satisfactory results can be obtained inmostcases irrespective of the detailed nature of the algorithmDependent on type of artifact and data to be cleaned how-ever a combination of approaches employing specific detec-tion metrics might be needed in order to reduce unwantedsignals to an acceptable level It is noteworthy to know thatvalidation has so far been performed in healthy volunteersonly making it difficult to judge when fully automatedmethods will meet clinical standards

42 Head Movement Head or more general subject move-ment during signal acquisition can significantly affect theusability of MEG data due nonneuronal signals arising frommuscle activity artifacts caused by motion related scannervibrations and changes in spatial reference impacting sub-sequent source localization Typically healthy adult patientsare highly motivated and maintain a constant head positionwithin about 2mm standard deviation of the measured headpositions in particular when scans are short Movementrelated artifacts however might increase substantially incertain patient populations and in pediatric measurements[4] Early approaches aimed at preventing head movement asfar as possible with deflatable support cushions bite-bars orother constraints In contrast modern approaches increasingrely on continuous position monitoring in conjunction with

computational methods to correct for head movementsduring data preprocessing

Arguably the most advanced approaches utilize signal-space-separation (SSS) This technique is based on funda-mental properties of quasi-static magnetic fields implyingthat the measurement geometry can be divided into twosource volumes with respect to the sensor array The internalvolume is a sphere whose radius is defined as the distancefrom the head origin to the nearest MEG sensorThe externalvolume is defined as the outside of a spherewhere the radius isthe distance from the head origin to the farthest MEG sensorThen themagnetic fieldmeasured by the sensors (119872) locatedin the current-free space between the internal and externalvolumes can be represented as

119872 = 119872int +119872ext = 119867 times 119860 = [119867int 119867ext] [119860 int119860ext

]

= 119867int times 119860 int + 119867ext times 119860ext

(8)

where 119867 depends on the sensor array and its location withrespect to the two source volumes In contrast the multipolemoments 119860 are device and geometry independent and canbe estimated from the original measurement 119872 Only theinternal terms in the formula above are of interest as they areassumed to represent sources in the brain and a signal-spaceseparated (filtered) signal is obtained by omitting the externalterms thereby suppressing artifacts originating far away fromthe MEG sensors Critically the device independency of 119860allows correction of head movements according to the fol-lowing

119872119904

= 119867119904

times 119860 int (9)

8 ISRN Radiology

Table 2 Overview of some of the relevant ICA-based methods for ocular andor cardiac artifact removal in chronological order

First author Artifact Electrode Sample Comment

Rong [65] Ocular cardiac No 5 One of the first automated removal methods requires a predeterminedartifact template

Okada [66] Ocular Yes 7 Reliable performance for blinks occurring in rapid successionMantini [67] Ocular cardiac No 12 Requires complex parameter tuning can correct environmental artifactsDammers [68] Ocular cardiac Yes 6 Uses signal amplitudes and phases can correct for muscle artifactsKlados [69] Ocular Yes 27 Hybrid regression-ICA approach available as open sourceElectrode indicates whether an electrooculogram andor electrocardiogram measurement is required Sample number of healthy volunteers used for valida-tion

where 119867119904

represents a virtual sensor array locked to thesubjectrsquos head In other words one can calculate the signalsrecorded by a sensor array with respect to where the head isstationary provided that a continuousmovementmonitoringmethod is available [71 72] A temporal extension of thisapproach known as tSSS attempts to identify residual signalscommon to 119872int and 119872ext Such signals typically but notnecessarily represent artifacts originating between the insideand outside volumes and once identified can be removedfrom119872int using a projection as described previously [73]

The SSS approach has been tested extensively wherethe suppression of artifacts originating at a large distance(gt1m) from the scanner matches the theory-based expec-tation [74 75] Note that suppression of nearfields such asocular and cardiac artifacts can be unsatisfactory Excellentresults however have been obtained for tSSS in the case ofspeech artifacts where the movements of tongue lips andjaw and the activity of facial muscles produce magnetic fieldsinterfering with the signals [76] This appears to hold alsofor magnetic fillings and other dental implants significantlyincreasing the number of cases that can be successfullyscanned with MEG

Although movement correction based on SSS has beenvalidated using artificial sources relatively few studies haveinvestigated the effects of head motion in vivo Initiallythe location error of the early somatosensory response toelectrical median nerve stimulation was calculated in a singlesubject making controlled headmovements [77]The authorsreported a 3-fold reduction in localization error comparedto raw data when using tSSS-based movement correction forhead shifts up to 3 cmThe limit of 3 cm is broadly consistentwith a single-case study reporting consistent localization ofepileptic activity when correcting for head movement duringseizure (23 arc length) whereas activity located in differentsublobar regions without correction [78] (Note the MEGresults did not change patientmanagement see Section 51 fora discussion of MEG in epilepsy research)

A recent study systematically investigated the effectsof head motion in a larger sample size [79] The authorsrecorded the neuronal response to auditory and somatosen-sory stimuli in 20 healthy volunteers performing controlledhead movements Moreover magnetized particles wereattached to the subjectrsquos head to simulate nearby interferencesources and a phantom head containing current dipoles wasused for cross validation Confirming and extending previousfindings the results showed that source locations can be

recovered reliably with SSS (error about 5mm) for headshifts not exceeding 3 cm Also using tSSS to remove artifactsdue to nearby sources did not affect the reference dataThe authors pointed out that auditory evoked responsesappeared to be more affected by continuous movement thansomatosensory responses implying that the outcomes ofmotion correction have to be judged in a context dependentmanner

43 Implanted Materials Implanted metal containing ob-jects such as electrodes hydrocephalus shunts or ferromag-netic remnants of neurosurgical operations pose particu-larly high challenges for MEG artifact correction as non-physiological magnetic fields are generated ldquoin siturdquo directlyinterfering with the brain signals The artifacts become evenworse when electrically active devices are involved such asbrain stimulators or pacemakers There is currently no uni-versally applicable method available that corrects for thesetypes of artifacts however promising results are beginningto emerge Two examples may suffice for illustration here

One study used MEG to record epileptic discharges inpatients carrying a vagus nerve stimulator which has provenbeneficial to some epileptic patients considered unsuitablecandidates for resective surgery [80] Using tSSS to clean thedata the authors were able to obtain interpretable signals in 7out of 10 patients The clean signals were subsequently mod-eled using equivalent dipoles in order to localize interictalspikes As a consequence of the MEG results managementchanged to invasive video-EEG (IVEM) in two patientsFurthermore the MEG findings supported the proposedmanagement in four patients demonstrating the feasibility ofMEG in patients with a nerve stimulator

Another study recorded simultaneous MEG and localfield potentials in 17 subjects with Parkinsonrsquos disease (PD)carrying a deep brain stimulator (DBS) to alleviate symptomscaused by degrading motor control [81] Using beamform-ing the authors were able to suppress the high-amplitudeartifacts caused by the DBS wire and electrode and extractinterpretable virtual electrode time-series Subsequent time-frequency analysis pointed to specific synchronization in thegamma band (60ndash90Hz) that stimulates movement throughmodulatory effects in the basal ganglia cortical networkAlthough not of direct clinical relevance this finding is ofimportance for a better understanding of motor control inPD

ISRN Radiology 9

5 Clinical Application Epilepsy

The most well-established clinical application of MEG ispresurgical evaluation in epilepsy Epilepsy is a diverse set ofneurological disorders characterized by the tendency to haverecurring seizures A seizure is the response to an abnormalelectrical discharge in the brain and it describes a varietyof behaviors such as psychic aberrations (eg a sense ofdeja vu) abnormal sensations (eg sensing an intense smell)impairments in speech or speech comprehensions and coor-dinated and involuntary movements Both the nature andseverity of seizure symptoms are dependent on which partsof the brain are affected by the abnormal discharges Seizuresof low severity leave consciousness unimpaired whereas themost severe seizuresmay become life-threatening and requireinstant medical attention [82 83]

Often it is unknown what causes abnormal electricaldischarges however certain risk factors have been identifiedsuch as brain lesions disorders of themetabolic system braininfections and drug and substance abuse amongst others Aseizure typically lasts up to several minutes implying that thetime period that exists between seizures (interictal period)accounts for the majority of a patientrsquos life with epilepsyAlthough most patients may appear symptom free duringinterictal periods abnormal dischargesmay be presentwithinthe brain The anatomic origin of such discharges is oftenalthough not always close to the anatomic origin of seizure-related (ictal) discharges implying a clinical relevance ofinterictal epileptic activity [84]

In industrialized countries epilepsy has an estimatedprevalence of 4ndash10 per 1000 and an incidence of 40ndash70 per100000 and year causing substantial medical expendituresThe disorder is usually controlled but not cured withantiepileptic drugs resulting in seizure freedom in about60 of patients The remaining 40 of affected individualshowever have to be considered pharmacoresistant [85]Surgery is an alternative option in these cases if the seizurescan be traced to abnormal electrical discharges in oneor more clearly delineated brain areas Then the removalor disconnection of the epileptogenic tissues can entail highrates of success Epilepsy surgery however faces substan-tial challenges The ultimate goal is improvement of qual-ity of life while avoiding cognitive or other impairmentscaused by damage to essential functional areas This requiresa substantial multitechnological and possibly invasive pre-operative workup aimed at both the accurate localization ofthe epileptic foci and mapping of nearby essential functionalareas where only the combined results can enable successfulsurgery [86]

51 Localization of Epileptic Foci The potential of MEGto localize epileptogenic regions was communicated to thewider scientific audience over 25 years ago [87] Efforts atconfirming the accuracy of MEG source localizations havebeen addressed from various indirect and direct approachesIndirect validation comes predominantly from studies show-ing colocalization of the magnetic source estimates withknown epileptogenic tissues apparent on structural and func-tional MR imaging or confirmed with invasive intracranial

EEG (IC-EEG) and successful surgical outcomes In contrastthe direct methods reflect studies utilizing simultaneousMEG and IC-EEG recordings or simulated epileptic dis-charges generated by implanted dipoles [85 88] Similar toother non-invasive approaches to preoperative evaluation aconsiderable amount of the MEG work has been based oninterictal discharges (an example is shown in Figure 3(a)[55 89ndash91]) However the importance to determine thelocation of the onset of ictal activity associated with clinicalseizures has prompted researcher to address methodologicalchallenges and validate ictal MEG [92]

In one of the first larger studies of its kind successfulictal MEG recordings were made in 6 out of 20 patientswith intractable complex partial seizures [94] All patientshad neocortical epilepsies and data collection took placeover a span of 3 years In order to capture ictal-onsetactivity a physician remained in the magnetically shieldedroom with each patient and retracted the probe soon after aseizure occurred allowing the patient to move freely Duringrecording a secondphysicianmonitored real-time displays ofthe MEG and EEG waveforms and triggered data collectionwhen abnormal discharges were detected On average theictal scans took several hours and the patients were allowedlight sleep unless it was necessary to reposition the head or ifsleep to wake transitions were felt necessary to obtain ictaldata This allowed the authors to record a few seconds ofictal data although their procedure had a limited capabilityto track seizure spread

The authors used single equivalent current dipole tolocalize sources underlying the MEG field patterns detectedduring ictal and interictal data segments Compared to IC-EEG the authors reported that ictalMEGprovided localizinginformation that was superior to interictal MEG in three ofthe six patients Localization of ictal onset byMEGwas foundto be equivalent to invasive EEG in five of the six patientsIn one patient ictal-intracranial EEG showed nonlocalizingdiffuse seizure related activity whereas the MEG resultspointed to a circumscribed ictal onset zone in right prefrontalcortices A resection was performed based in part on theMEG findings leaving the patient seizure free and withoutpostoperative neurocognitive deficit over 2 years of followupDespite a small number of ictal recordings the authors rightlyconcluded that ictal MEG might be superior to invasiverecordings under certain conditions

The study furthered two previous investigations in 4patients [95] and 3 patients [92] with epilepsy respectivelyBoth studies concluded that the ictal MEG results were inagreementwith invasive andnoninvasive preoperative assess-ments Also the ictal and interictal MEG source localizationswere topographically very similar These findings howeverwere rather preliminary in nature and insufficient informa-tion was available as to assess the reliability of the utility ofMEG in identification of epileptogenic zones

Further support for the role of ictal MEG in presurgicalevaluation is provided by a recent study in children withintractable complex partial epilepsy [96] Over a span of35 years the authors successfully obtained ictal data in8 pediatric patients who subsequently underwent surgicalresection During MEG recordings the patients were either

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

2 ISRN Radiology

established implying a reasonably defined role of MEG indiagnosis treatment and patient management whereas thelatter is an emerging field forMEG reflecting a growing trendin using neuroimaging in the study of developmental disor-ders

2 MEG Technology

MEG is based on the detection of the magnetic fields that aregenerated by the currents flowing in neurons Unlike positronemission tomography (PET) or functional magnetic reso-nance imaging (fMRI) MEG does not rely on secondaryeffects induced by brain activity but directly measures themagnetic fields primarily generated by postsynaptic neuronalionic currents In contrast to its closest cousin the elec-troencephalogram (EEG) MEG signals are reference-freeand essentially unaffected by conductivity differences on themagnetic flux providing an almost undistorted view of brainactivity and simplifying data analysis and interpretation ofpatterns MEG is mainly sensitive to magnetic fields gener-ated in the cerebral cortex but modern whole-head systemswith multiple sensor configurations can detect activity insubcortical regions Typically neural magnetic field changesare extremely small ranging from 10minus14 T or even lessfor evoked fields to approximately 10minus12 T during interictalepileptic spikes [21 22] By comparison a typical fMRI scan-ner operates at 15 T and the Earthrsquos magnetic field is approx-imately 10minus4 T Currently the only practical and competitivedetectors capable of recording such small field changes aresuperconducting loops that are coupled into superconductingquantum interference devices (SQUID) that respond to thechanges inmagnetic flux through the loops [23 24] AnMEGscanner is illustrated in Figure 1

The loop configuration determines what the detectormeasures for example a simple planar coil measures thechange in the magnetic field perpendicular to the loop planeand a figure-of-eight-shaped coil measures the change in themagnetic field gradient An analogy is in orderThe loop (coil)can be compared to a lens forming an image dependent on thefocal length where the SQUID transforms themagnetic fieldsinto electrical voltages much like a camera sensor transformsan optical image A typical MEG imaging system has severalhundred sensors surrounding a head shaped recess in a liquidhelium-filled Dewar (cryogenic temperatures are requiredfor SQUID operation) The whole instrument is housed ina magnetically screened room that is required to reduce theambient magnetic field noise The best instruments currentlyavailable achieve noise levels approaching 10minus16 T per squareroot of the bandwidth over which the signals are recordedThe temporal resolution of MEG is less than a millisecondwhich is sufficient to trace even the fastest neuronal processesThe spatial resolution of MEG depends on signal quality andcan be as low as a few millimeters for appropriate sourcemodels which is in the range of subdural electrode strips andgrids [6 25]

There are subtleties associated with the sources that aredetected Even with modern scanners the fields emanatingfrom single nerve cells are too weak to be measured outsidethe head Similar to EEG this implies that detectable field

changes require large numbers (few thousand to a few tenthousands) of neighbouring cells to carry aligned currentsof suitable strength and duration to allow for spatial andtemporal and spatial summations It is commonly assumedthat these conditions are met through postsynaptic currentsrather than currents associatedwith action potentials or otherelectrobiochemical processes [26ndash28] and [29 Chapter 1]MEG is technically challenging however this technologyhas some significant advantages It is a passive and con-tactless measurement that is completely non-invasive easyto prepare and minimally demanding on the subject Theinstrument is silent in its operation where the MEG detectorrecords the brain activity continuously as the subject per-forms tasks required by the investigator Longitudinal studiesare ethically possible MEG is capable of providing meaning-ful results on its own although information extracted fromother brain imaging techniques and coregistered with theMEG can be used to guide analysis and interpretation

3 Analytical Approaches

The primary outputs provided by any magnetoencephalo-graphic scanner are readings of the variation in time of themagnetic fields produced by electrical activity in the brainand detected over the head These signals are analyzed invarious ways correlated with experimental conditions andinterpreted within the framework of a given clinical inves-tigation where often findings obtained from other imagingmodalities are considered or even integrated into the analysisA plethora of analytical techniques to extract informationfrom the MEG signals have been developed and new meth-ods continue to emerge rapidly Currently the abundance ofapproaches implies an absence of standard protocols forMEGanalysis but there is a clear drive towards standardizationand guideline papers are beginning to emerge [30] Despitethe variety and partial incomparability of methodologieshowever certain approaches have crystallized that are usedfairly consistently across clinical studies

31 Event-Related Fields Event-related fields (ERF) are directanalogues of the perhaps more familiar event-related poten-tials (ERP) Event-related approaches rely on repeatedlypresented auditory visual tactile electrical or other stimuliThe appearance of each stimulus marks a point in timeusually denoted as stimulus onset with respect to which anepoch of data is defined The signal observed in the post-stimulus interval is assumed to be independent of stimulusrepetition but to be contaminated by random brain signalsconsidered to be noise This assumption is often justified butmay have limited validity in studies of higher order functionWithin the assumption of independence against repetitionthe stimulus-specific neural response can be recovered fromthe measured signal by averaging the data with respect tostimulus onset The resultant waveform is usually called anevoked response The process of averaging may be extendedover several subjects to yield grand mean waveforms Typi-cally the stimulus and subject dependence of peak amplitudesand latencies of the evoked responses allow some insight intotheir functional significance [31]

ISRN Radiology 3

Elektra NeuromagTM

3 seconds

600 f

Tcm

Figure 1 The MEG scanner at the Oxford Centre for Human Brain Activity Left Shown is a typical set up where the subject is seatedupright under the scanner A button box is used to record behavioral responses and an eye-tracker supplies additional psychometrical dataMeasurements in supine position are possible and a doctor or nurse can stay within the magnetically shielded room in order to monitor thepatient if needed Middle The brain magnetic fields are recorded with a helmet-shaped array of 102 SQUID devices featuring 3 pick-up coilseach (306 channels in total) Two first-order gradiometers measure two orthogonal spatial gradients of the magnetic field in longitudinal andlatitudinal directions respectively These channels are most sensitive to the tangential neuronal currents in the region below the device Amagnetometer coil (black rectangle) measures the magnetic field and is sensitive to deeper sources Right Typically the cortex-to-detectordistance is 4-5 cm The primary output of the scanner is a trace for each channel representing the magnetic field or gradient as a function oftime (for presentation only 16 channels are shown brain image courtesy of Elekta Neuromag OY Helsinki)

32 Time-Frequency Analysis Viewing brain activity as anoscillatory phenomena there is a long history of studyingfundamental brain rhythms such as theta (4ndash7Hz) or alpha(8ndash13Hz) waves that under certain conditions may indicategeneral states of the brain for example increased alpha isoften seen when subjects close their eyes [32] Robust algo-rithms are now available to analyze neurophysiological sig-nals in terms of oscillations that vary in frequency and timeTypically transient oscillatory brain activity is consideredin the context of well-defined stimuli and is identified aseither an evoked oscillatory response appearing at the samelatency and phase in each trial or an induced oscillatoryresponse appearing with a jitter in phase from one trial toanother and therefore not detectable in the averaged dataBoth types of response are assumed to provide informationabout brain dynamics beyond what can be extracted from theconventional evoked response which includes all frequencies[33 34]

Oscillations at higher frequencies in the beta (15ndash30Hz)and gamma (gt30Hz) bands have received much attention inrecent years Experimental data reveal that synchronizationof activity in these bands is involved in processes such asworking memory semantic processing and sensory-motorintegration and that synchronization may be altered underpathological conditions [35 36] Theoretical considerationssuggest that synchrony of oscillations can facilitate thecoordinated interactions of large neuronal population dis-tributedwithin and across distinct regions of the brain where

interneuron networks are believed to play an important role[37 38]

Of particular relevance to clinical application is event-related desynchronization (ERD) ERD refers to the well-known phenomenon that certain events or stimuli can desyn-chronize or even block parts of the ongoing brain activityThese induced changes can be detected using a techniqueemploying band-pass filtering squaring and averaging inorder to obtain a measure of the oscillatory power in a givenfrequency band of interest ERD is then quantified relativeto the oscillatory power obtained from a reference intervalThe ERD measure may become positive dependent on thenature of task and stimuli in which case one uses the termevent-related synchronization (ERS) ERD (ERS) may beconsidered to be due to a decrease (increase) in synchronyof the underlying neuronal populations and is recognized asa robust marker of normal and abnormal neuronal processesin clinical applications of EEG and MEG [39]

33 Source Estimation Most relevant to clinical MEG issource estimation that is the identification of the brain areasgenerating a given signal recorded by the detectors (alsoknown as magnetic source imaging MSI) Source estimationis ultimately based on the theory of electromagnetism whichis governed by 4 equations known as Maxwellrsquos equationswhich relate the spatial and timederivatives of the electric andmagnetic fields to the charges and currents present within agiven volume defined by its geometry electric permittivity

4 ISRN Radiology

and magnet permeability [40] In the case of electrophysi-ological signals the useful frequency spectrum is typicallybelow 2 kHz and most studies deal with frequency between005Hz and a few hundred hertz As a consequence onecan disregard the time derivatives yielding the quasi-staticapproximation

119861 (119903) sim int119871 (119903 1199031015840

) times 119869 (1199031015840

) 1198891199031015840

(1)

also known as Biot-Savartrsquos law The rule relates the currentflow 119869 at location 1199031015840 within a given source volume to the mag-netic field 119861 observed at position 119903 The integral (continuoussum) runs over the volume containing the currents where thelead field 119871 establishes a link between currents and magneticfield taking into account volume geometry and propertiesTypically one associates the magnetic field with the actualmeasurement in which case the lead field is also dependenton the type position and orientation of the detector usedThen the physical interpretation of the lead field is that itrepresents the sensitivity pattern of a given detector and agiven source volume [21 41]

Conceptually the law constitutes a forward problem thatis knowing the currents and the properties of the embeddingmedium one can calculate the magnetic field in an analyticalor numerical fashion In contrast every source estimationprocedure has to face the so-called inverse problem that isthemathematical transformation that uses the observedmag-netic field as input and determines the sources of the signalsHere there is a problem that is shared with many algo-rithmic proceduresmdashthe inverse problem in MEG is deeplymathematically ambiguous [41 42] No matter how highquality the data is and howmany measurement channels oneuses there can always be silent albeit neurophysiologicallyrelevant currents in the brain that are not recorded by thedetectors The MEG inverse problem is approached prag-matically by simplifying the description of the sources orby adding additional physiologically reasonable assumptionsThe forward model is always needed for source estimationand it is commonly assumed that the current flow in thehead volume can be expressed as the sum of a primary orimpressed current due to the electromotive forces associatedwith biologicalmainly synaptic activity in neural tissue and apassive current flow that results from motion of extracellularcharge carriers [43] Consider

119869 (119903) = 119869impressed (119903) + 119869passive (119903) (2)

331 Equivalent Current Dipoles The arguably simplest anduntil recently most commonly used source estimation tech-nique is the use of equivalent current dipoles (ECD) An ECDcan be viewed as the spatial average of all impressed currentswithin an area of the brain assuming that the arearsquos spatialextent is small compared with the distances to the detectorsEach equivalent dipole is characterized by 6 parametersdescribing the dipolersquos position orientation and strengthwithin the head In general a nonlinear iterative algorithmis applied in order to find the parameters best describingthe data measured In each iteration step the magnetic fields

generated by the dipoles are compared with the fields mea-sured by the detectors and the parameters are adjusted tominimize the field difference quantified by a cost functionTypically a single homogenous sphere is used to approximatethe head volume although other conductormodels have beenused such as ellipsoids several spheres each fitted to thelocal curvature of the head or realistically shaped geometriesbased on individual structural scans

In case of the homogenous sphere the magnetic forwardproblem becomes independent of tissue conductivities andthe spherical symmetries imply that radially oriented currentdipoles become hidden Thus this source model can onlyaccount for tangential dipoles assumed to be primarilylocated in the walls of cortical fissures and sulci but not in theconvexial cortex [21 41 44] ECD approaches vary accordingto the number of dipoles employed (one dipole being themost common choice) the number and type of parametersallowed to change during fit (eg fixed position with rotatingorientation and time-varying strength) and the nature ofsearching or scanning algorithm used to fit parameters todata

332 Imaging Methods Imaging approaches to the inverseproblem explore quasi-continuous estimates of activation bydividing the brain volume into distinct voxels and allocating3 orthogonal equivalent current dipoles to each voxel Severalthousand voxels are usually employed approximating anyarbitrary spatial distribution of post-synaptic currents in thebrain The inverse problem in this case is linear since thesource locations are known leaving only source amplitudes tobe estimated The problem however is severely underdeter-mined given that the number of detectors is small comparedto the number of voxels and regularization is required inorder to restrict the range of allowable source currents whilesimultaneously minimizing the difference betweenmeasuredand predicted magnetic fields

Symbolically

minimum119869

119861 minus 119871 times 119869119901 + 1205821003817100381710038171003817119882model times 119869

1003817100381710038171003817

119901

(3)

where sdot sdot sdot 119901 denotes a norm (roughly strength) of thequantity enclosed In other words one estimates for all voxelsthose current amplitudes that best explain the data and keepa chosen model term also known as a penalty term smallCommonly the voxels are not uniform throughout the cra-nium but restricted to the grey matter or more specifically tothe cerebral cortex In line with the assumption that currentsin the dendritic trunks of pyramidal cells predominantly con-tribute to external fields the dipole orientation may furtherbe constrained to be perpendicular to the local cortical sur-face The parameter 120582 controls the amount of regularizationand its optimal value is usually determined in a data-drivenfashion using an appropriate heuristic approach The weightmatrix 119882model is used to assign additional properties of theinverse solution such as focal sources In essence the weightand the normparameter119901 together define the type of imagingsolution each of which has its merits and disadvantages [44ndash50] The most common choices of norms and weights aresummarized in Table 1

ISRN Radiology 5

Table 1 Comparison of distributed source algorithms often used in clinical MEG research

Name 119901 119882model Comment

MNE 2 1+ simpleminus currents packed towards surface of volume

MNE (weighted) 2 1current-gain + no surface biasminus blurry images

LORETA 2 Laplacian+ + deep source accurateminus estimates superficial sources too deep

MCE 1 1current-gain + focal estimatesminus strong sensitivity to noise

MNE minimum norm estimate (eg [42]) MCE minimum current estimate [46] LORETA low resolution brain electromagnetic tomography [45] +Math-ematical operator representing flux density and gradient flow of a function

333 Beamforming Beamforming approaches to the inverseproblem aim at discriminating between signals originatingfrom a location of interest and those generated elsewhere inthe brain Similar to imaging techniques one assumes thatthe MEG measurement can be modeled by a set of fixedcurrent dipoles distributed throughout the brain In contrastto imaging techniques however one does not attempt to solvea set of underdetermined equations using norm constraintsInstead a beamformer estimates the neural activation at eachvoxel independent of all other locations by means of a spatialfilter Ideally such filter 119865 has complete spatial specificityzeroing the lead field at all locations except the chosen one

119865 (1199030

) times 119871 (119903 1199031015840

) =

1 1199031015840

= 1199030

0 1199031015840

= 1199030

(4)

where 1199030

and 1199031015840 are locations with the source volume and119903 refers to the location of measurement Once the filter hasbeen estimated the neuronal current at position 119903

0

can berecovered from the magnetic fields detected by the sensorarray

119869 (1199030

) = 119865 (1199030

) times 119861 (5)

This expression is commonly interpreted as a virtual electrodethat is one gains knowledge about a current as if an electrodewas attached to the location of interest In practice it isgenerally impossible to have complete attenuation in the stop-band that is locations other than the one of interest and oneaims for designing a filter that is optimal in some sense[44 51]

Arguably the most influential guiding philosophy fordesigning an optimal filter is the linear constrainedminimumvariance approach or LCMV for short [51] The idea is todetermine a filter that minimizes the variance of the sourceat a given location while satisfying the linear response con-straint 119865 times 119871 = 1 ensuring that the signals of interest passby the filter Minimization of source variance which is ameasure of current strength optimally allocates the stop-band response of the filter tominimize the contribution to theoutput due to signals generated by strong sources in the stop-band The mathematical apparatus of beamforming relies onthe assumption that the dipole sources generate mutuallyuncorrelated brain activity Correlated activation cannot bereliably disambiguated from correlated signals at the sensor

level due to volume conduction and field propagation Inother words the sensor readings might be correlated becauseunderlying activations are correlated or the field generated bya source is detected bymore than one sensor at the same timeThe assumption that brain activity is uncorrelated acrossregions might not be physiologically justifiable however ithas been shown that beamformer source estimates are tosome extent robust against correlated sources [44 51 52]

A related approach is known as hardware beamformingexploiting the physical properties of certain detectors typesSensors like planar gradiometers employ two coils measuringorthogonalmagnetic field gradients (spatial variation) at eachsensor location The root-mean-square of the two readingscan be interpreted as the strength of a current dipole locateddirectly below each gradiometer yielding a sensor-level mapof brain activation that can be projected onto the corticalsurface The method is easy to implement and might beapplicable even if the scanner does not have sensors directlysuitable for hardware beamforming [53]

334 Summary Source Estimation The researchers and clin-icians can now draw on awide range of source reconstructionmethods often available in form of commercial and opensoftware packagesThe source images obtained fromdifferentalgorithms are often broadly consistent and may in favor-able circumstances offer meaningful insights into functionalanatomy at the centimeter level However no general sourceimaging method is currently available where the results maydepend on the data and method of analysis In particular thespatial extent of source images may reflect the properties ofthe estimate rather than the true extent of the neural currentStatistical inferences based on source estimates need to bejudged with care as significance does not necessarily implysignificantly nonzero real primary current in a voxel [54] Inany case the interpretation of source estimates should includeconsideration of the nature of the underlying algorithm

4 Artifacts and Artifact Removal

The high sensitivity of MEG implies that the recordings aresensitive to artifacts originating from a variety of physiolog-ical and environmental sources This issue is not trivial andcan become a serious challenge in certain patient populationsdue to amongst others heightened physiological responses

6 ISRN Radiology

lack of control over movements and magnetic materials nec-essary for diagnosis and treatment [4] In general the impactof artifacts may be reduced or controlled to some extentthrough careful preparation and experimental design andexclusion of the most affected data may facilitate analysis andinterpretation However it is impossible to exploit clinicaldata fully without advanced approaches to signal processingthat can reliably separate brain signals and artifacts (Figure 2illustrates the concepts discussed in this section)

41 Physiological Artifacts Arguably the two most dominantartifact sources are the heartbeat and eye blink Althoughthese physiological artifacts can have rather large amplitudessignal separation is challenging as the artificial signals mayresemble activity originating from neuronal sources Regard-ing the eye normal automatic blinking occurs between about5 and 20 times per minute and causes magnetic fields overpredominately frontotemporal regions that can be muchlarger than the magnetic fields of spontaneous and evokedbrain activity The primary artifact is generated by currentflow between the eye and scalp and by excitation of ocularmuscles controlling eye and eyelid movements The artifacttime course recorded by MEG or EEG during a spontaneousblink corresponds to the eyelid motion reaching a closedposition on average 120ms after the blink is initiated andbefore returning to a fully open position approximately240ms after closure In addition to the direct artifact fieldsblink related neuronal activity lasting up to several tenths ofa second is also often observed over posterior parietal andfrontal cortices as well as other areas of the brain associatedwith higher-order processing [57 58] For artifact correctionone generally obtains an independent record of the artifactby measuring the electrooculogram (EOG) with electrodesplaced close to the eyes

Regarding the heart electrical currents moving throughthe organ during a heartbeat give rise to magnetic fields thatare over the chest between two and three orders of magni-tude larger than the brain magnetic fields The time course ofthe cardiac artifact closely corresponds to the propagation ofelectrical activity through the heart muscle Each heartbeatbegins with an impulse from the sinoatrial node the heartrsquosmain pacemaker activating the upper chambers (atria) Thisactivation is seen as the P wave (duration about 80ms) in theelectrocardiogram (ECG) Next the electrical current flowsdown the heart and activates the lower chambers (ventricles)giving rise to a characteristicwave known as theQRS complex(80ndash120ms) Subsequently the electrical current spreadsback over the ventricles in the opposite direction generatinga recovery wave commonly denoted by the letter T (160ms)Typically these currents flowing in the heart are observableto varying degrees in MEG and EEG recordings causingartifacts that may resemble epileptic spikes and slow waves Itis generally assumed that secondary artifacts due to heart beatrelated blood pulsation body movements (ballistocardio-effects) and susceptibility changes are negligible but mayexist in MEG and EEG recordings if the lungs skin or bloodvessels contain magnetic particles [59 page 162] and [60]For artifact correction one generally obtains an independent

record of the artifact by measuring the ECG with electrodesplaced on both arms close to the wrists

Various approaches have been suggested in the past tocorrect for these artifacts often but not exclusively utilizingequivalent current dipoles to model nonneuronal aspectsof the artifacts for example the retinarsquos electrical charge[61] Despite merits however those methods are increas-ingly being replaced by signal projection-based approachesProjection methods are broadly analogous to obtaining a2-dimensional shadow from a 3-dimensional object wherethe direction of illumination determines which features arevisible in the projection Mathematically the data are repre-sented as points (also called vectors) in a space of dimension119899 equal to the number of measurement channels Each 119899-dimensional vector corresponds to one sampled time pointThen the set of new vectors 119881 is calculated to form a basisfor the unwanted signal features that is the artifact to beremoved In other words the set 119881 spans a subspace that isas far as possible occupied by the artifact but not the neuro-physiological signals of interest An artifact-free (clean)measurement is readily obtained through projection

119872119888

= (1 minus 119881 times 119881+

) times 119872 (6)

where 119872 denotes the measurement matrix of dimension119899 times (number of time-points) 1 denotes the 119899-dimensionalidentity matrix and 119881+ is the pseudoinverse of 119881 which hasdimension 119899 times 119896 where 119896 is less than 119873 [62] Dependenton the artifact to be removed 119896 typically assumes a valuebetween 1 and 3

Perhaps the most powerful method to determine theartifact subspace 119881 is a type of blind source separation(BSS) known as independent component analysis (ICA) ICAis entirely data driven and assumes that the observations119872 constitute a linear superposition (mixing) of statisticallyindependent source signals The number of independentsources 119904 is assumed to be the most equal to the number ofmeasurement channels A frequently quoted analogy is theldquococktail party problemrdquo In this scenario party guests holdmultiple simultaneous conversations recorded using micro-phones located at fixed positions around the room Knowingthe number of speakers (s le number of microphones) ICAallows recovering of the individual conversations by unmix-ing the recordings

119878 = 119880 times119872 (7)

where the number of rows of 119878 is equal to the number ofspeakers Each row of 119878 contains the sound from one speakeronly The number of columns of the un-mixing matrix 119880 isequal to the number of microphones and each element of119880 can be interpreted as a weight determining how much anindividual speakerrsquos voice contributed to a given microphone[63 64] Using this analogy the task is then to identify theldquospeakersrdquo who deliver the artifact and guide them out theroom

Over the last 7 years or so several algorithms have beenpublished in order to address this issue in an automatedfashion The main feature that distinguishes the differentapproaches is whether or not an independent EOG andor

ISRN Radiology 7

EOG

CorrectedOriginal

Source power

Original Corrected

Front

Back

RightLeftSi

gnal

(fT

cm) 300

200100

0minus100

minus200

Time (s)4 5 6 7 8

(a)

(b)

(c)

Figure 2 Artifact removal (a) Shown are data from one channel before (black) and after (red) ICA-based correction for eye-blink artifactsidentified by EOG activity (bottom trace) The inset illustrates the 2-dimensional artifact space used for projection (time-courses and spatialtopographies) The first dimension is dominant Note the artifact has similar features as epileptic discharges in that it involves several MEGchannels has sharp peaks and stands out from ongoing background activity (see also [55]) (b) Equivalent current dipole modeling of anevoked somatosensory response in a child before (left) and after (right) SSS-based movement correction After correction the ECD assumesa physiologically plausible location (courtesy of Elekta Neuromag OY Helsinki) (c) Brain activity can be localized accurately despite strongartifacts caused by DBS electrodes (right) Before beamforming-based correction strong nonphysiological interferences outside the brainare observed (left adapted from [56])

ECG recording is needed for artifact identification Table 2shows a brief overview of some of the relevant studies

A recent study in 26 healthy volunteers compared variousartifact detection metrics as well as different flavors of ICAand other blind source separation techniques [70] In gen-eral the authors found automated artifact removal methodspracticable where satisfactory results can be obtained inmostcases irrespective of the detailed nature of the algorithmDependent on type of artifact and data to be cleaned how-ever a combination of approaches employing specific detec-tion metrics might be needed in order to reduce unwantedsignals to an acceptable level It is noteworthy to know thatvalidation has so far been performed in healthy volunteersonly making it difficult to judge when fully automatedmethods will meet clinical standards

42 Head Movement Head or more general subject move-ment during signal acquisition can significantly affect theusability of MEG data due nonneuronal signals arising frommuscle activity artifacts caused by motion related scannervibrations and changes in spatial reference impacting sub-sequent source localization Typically healthy adult patientsare highly motivated and maintain a constant head positionwithin about 2mm standard deviation of the measured headpositions in particular when scans are short Movementrelated artifacts however might increase substantially incertain patient populations and in pediatric measurements[4] Early approaches aimed at preventing head movement asfar as possible with deflatable support cushions bite-bars orother constraints In contrast modern approaches increasingrely on continuous position monitoring in conjunction with

computational methods to correct for head movementsduring data preprocessing

Arguably the most advanced approaches utilize signal-space-separation (SSS) This technique is based on funda-mental properties of quasi-static magnetic fields implyingthat the measurement geometry can be divided into twosource volumes with respect to the sensor array The internalvolume is a sphere whose radius is defined as the distancefrom the head origin to the nearest MEG sensorThe externalvolume is defined as the outside of a spherewhere the radius isthe distance from the head origin to the farthest MEG sensorThen themagnetic fieldmeasured by the sensors (119872) locatedin the current-free space between the internal and externalvolumes can be represented as

119872 = 119872int +119872ext = 119867 times 119860 = [119867int 119867ext] [119860 int119860ext

]

= 119867int times 119860 int + 119867ext times 119860ext

(8)

where 119867 depends on the sensor array and its location withrespect to the two source volumes In contrast the multipolemoments 119860 are device and geometry independent and canbe estimated from the original measurement 119872 Only theinternal terms in the formula above are of interest as they areassumed to represent sources in the brain and a signal-spaceseparated (filtered) signal is obtained by omitting the externalterms thereby suppressing artifacts originating far away fromthe MEG sensors Critically the device independency of 119860allows correction of head movements according to the fol-lowing

119872119904

= 119867119904

times 119860 int (9)

8 ISRN Radiology

Table 2 Overview of some of the relevant ICA-based methods for ocular andor cardiac artifact removal in chronological order

First author Artifact Electrode Sample Comment

Rong [65] Ocular cardiac No 5 One of the first automated removal methods requires a predeterminedartifact template

Okada [66] Ocular Yes 7 Reliable performance for blinks occurring in rapid successionMantini [67] Ocular cardiac No 12 Requires complex parameter tuning can correct environmental artifactsDammers [68] Ocular cardiac Yes 6 Uses signal amplitudes and phases can correct for muscle artifactsKlados [69] Ocular Yes 27 Hybrid regression-ICA approach available as open sourceElectrode indicates whether an electrooculogram andor electrocardiogram measurement is required Sample number of healthy volunteers used for valida-tion

where 119867119904

represents a virtual sensor array locked to thesubjectrsquos head In other words one can calculate the signalsrecorded by a sensor array with respect to where the head isstationary provided that a continuousmovementmonitoringmethod is available [71 72] A temporal extension of thisapproach known as tSSS attempts to identify residual signalscommon to 119872int and 119872ext Such signals typically but notnecessarily represent artifacts originating between the insideand outside volumes and once identified can be removedfrom119872int using a projection as described previously [73]

The SSS approach has been tested extensively wherethe suppression of artifacts originating at a large distance(gt1m) from the scanner matches the theory-based expec-tation [74 75] Note that suppression of nearfields such asocular and cardiac artifacts can be unsatisfactory Excellentresults however have been obtained for tSSS in the case ofspeech artifacts where the movements of tongue lips andjaw and the activity of facial muscles produce magnetic fieldsinterfering with the signals [76] This appears to hold alsofor magnetic fillings and other dental implants significantlyincreasing the number of cases that can be successfullyscanned with MEG

Although movement correction based on SSS has beenvalidated using artificial sources relatively few studies haveinvestigated the effects of head motion in vivo Initiallythe location error of the early somatosensory response toelectrical median nerve stimulation was calculated in a singlesubject making controlled headmovements [77]The authorsreported a 3-fold reduction in localization error comparedto raw data when using tSSS-based movement correction forhead shifts up to 3 cmThe limit of 3 cm is broadly consistentwith a single-case study reporting consistent localization ofepileptic activity when correcting for head movement duringseizure (23 arc length) whereas activity located in differentsublobar regions without correction [78] (Note the MEGresults did not change patientmanagement see Section 51 fora discussion of MEG in epilepsy research)

A recent study systematically investigated the effectsof head motion in a larger sample size [79] The authorsrecorded the neuronal response to auditory and somatosen-sory stimuli in 20 healthy volunteers performing controlledhead movements Moreover magnetized particles wereattached to the subjectrsquos head to simulate nearby interferencesources and a phantom head containing current dipoles wasused for cross validation Confirming and extending previousfindings the results showed that source locations can be

recovered reliably with SSS (error about 5mm) for headshifts not exceeding 3 cm Also using tSSS to remove artifactsdue to nearby sources did not affect the reference dataThe authors pointed out that auditory evoked responsesappeared to be more affected by continuous movement thansomatosensory responses implying that the outcomes ofmotion correction have to be judged in a context dependentmanner

43 Implanted Materials Implanted metal containing ob-jects such as electrodes hydrocephalus shunts or ferromag-netic remnants of neurosurgical operations pose particu-larly high challenges for MEG artifact correction as non-physiological magnetic fields are generated ldquoin siturdquo directlyinterfering with the brain signals The artifacts become evenworse when electrically active devices are involved such asbrain stimulators or pacemakers There is currently no uni-versally applicable method available that corrects for thesetypes of artifacts however promising results are beginningto emerge Two examples may suffice for illustration here

One study used MEG to record epileptic discharges inpatients carrying a vagus nerve stimulator which has provenbeneficial to some epileptic patients considered unsuitablecandidates for resective surgery [80] Using tSSS to clean thedata the authors were able to obtain interpretable signals in 7out of 10 patients The clean signals were subsequently mod-eled using equivalent dipoles in order to localize interictalspikes As a consequence of the MEG results managementchanged to invasive video-EEG (IVEM) in two patientsFurthermore the MEG findings supported the proposedmanagement in four patients demonstrating the feasibility ofMEG in patients with a nerve stimulator

Another study recorded simultaneous MEG and localfield potentials in 17 subjects with Parkinsonrsquos disease (PD)carrying a deep brain stimulator (DBS) to alleviate symptomscaused by degrading motor control [81] Using beamform-ing the authors were able to suppress the high-amplitudeartifacts caused by the DBS wire and electrode and extractinterpretable virtual electrode time-series Subsequent time-frequency analysis pointed to specific synchronization in thegamma band (60ndash90Hz) that stimulates movement throughmodulatory effects in the basal ganglia cortical networkAlthough not of direct clinical relevance this finding is ofimportance for a better understanding of motor control inPD

ISRN Radiology 9

5 Clinical Application Epilepsy

The most well-established clinical application of MEG ispresurgical evaluation in epilepsy Epilepsy is a diverse set ofneurological disorders characterized by the tendency to haverecurring seizures A seizure is the response to an abnormalelectrical discharge in the brain and it describes a varietyof behaviors such as psychic aberrations (eg a sense ofdeja vu) abnormal sensations (eg sensing an intense smell)impairments in speech or speech comprehensions and coor-dinated and involuntary movements Both the nature andseverity of seizure symptoms are dependent on which partsof the brain are affected by the abnormal discharges Seizuresof low severity leave consciousness unimpaired whereas themost severe seizuresmay become life-threatening and requireinstant medical attention [82 83]

Often it is unknown what causes abnormal electricaldischarges however certain risk factors have been identifiedsuch as brain lesions disorders of themetabolic system braininfections and drug and substance abuse amongst others Aseizure typically lasts up to several minutes implying that thetime period that exists between seizures (interictal period)accounts for the majority of a patientrsquos life with epilepsyAlthough most patients may appear symptom free duringinterictal periods abnormal dischargesmay be presentwithinthe brain The anatomic origin of such discharges is oftenalthough not always close to the anatomic origin of seizure-related (ictal) discharges implying a clinical relevance ofinterictal epileptic activity [84]

In industrialized countries epilepsy has an estimatedprevalence of 4ndash10 per 1000 and an incidence of 40ndash70 per100000 and year causing substantial medical expendituresThe disorder is usually controlled but not cured withantiepileptic drugs resulting in seizure freedom in about60 of patients The remaining 40 of affected individualshowever have to be considered pharmacoresistant [85]Surgery is an alternative option in these cases if the seizurescan be traced to abnormal electrical discharges in oneor more clearly delineated brain areas Then the removalor disconnection of the epileptogenic tissues can entail highrates of success Epilepsy surgery however faces substan-tial challenges The ultimate goal is improvement of qual-ity of life while avoiding cognitive or other impairmentscaused by damage to essential functional areas This requiresa substantial multitechnological and possibly invasive pre-operative workup aimed at both the accurate localization ofthe epileptic foci and mapping of nearby essential functionalareas where only the combined results can enable successfulsurgery [86]

51 Localization of Epileptic Foci The potential of MEGto localize epileptogenic regions was communicated to thewider scientific audience over 25 years ago [87] Efforts atconfirming the accuracy of MEG source localizations havebeen addressed from various indirect and direct approachesIndirect validation comes predominantly from studies show-ing colocalization of the magnetic source estimates withknown epileptogenic tissues apparent on structural and func-tional MR imaging or confirmed with invasive intracranial

EEG (IC-EEG) and successful surgical outcomes In contrastthe direct methods reflect studies utilizing simultaneousMEG and IC-EEG recordings or simulated epileptic dis-charges generated by implanted dipoles [85 88] Similar toother non-invasive approaches to preoperative evaluation aconsiderable amount of the MEG work has been based oninterictal discharges (an example is shown in Figure 3(a)[55 89ndash91]) However the importance to determine thelocation of the onset of ictal activity associated with clinicalseizures has prompted researcher to address methodologicalchallenges and validate ictal MEG [92]

In one of the first larger studies of its kind successfulictal MEG recordings were made in 6 out of 20 patientswith intractable complex partial seizures [94] All patientshad neocortical epilepsies and data collection took placeover a span of 3 years In order to capture ictal-onsetactivity a physician remained in the magnetically shieldedroom with each patient and retracted the probe soon after aseizure occurred allowing the patient to move freely Duringrecording a secondphysicianmonitored real-time displays ofthe MEG and EEG waveforms and triggered data collectionwhen abnormal discharges were detected On average theictal scans took several hours and the patients were allowedlight sleep unless it was necessary to reposition the head or ifsleep to wake transitions were felt necessary to obtain ictaldata This allowed the authors to record a few seconds ofictal data although their procedure had a limited capabilityto track seizure spread

The authors used single equivalent current dipole tolocalize sources underlying the MEG field patterns detectedduring ictal and interictal data segments Compared to IC-EEG the authors reported that ictalMEGprovided localizinginformation that was superior to interictal MEG in three ofthe six patients Localization of ictal onset byMEGwas foundto be equivalent to invasive EEG in five of the six patientsIn one patient ictal-intracranial EEG showed nonlocalizingdiffuse seizure related activity whereas the MEG resultspointed to a circumscribed ictal onset zone in right prefrontalcortices A resection was performed based in part on theMEG findings leaving the patient seizure free and withoutpostoperative neurocognitive deficit over 2 years of followupDespite a small number of ictal recordings the authors rightlyconcluded that ictal MEG might be superior to invasiverecordings under certain conditions

The study furthered two previous investigations in 4patients [95] and 3 patients [92] with epilepsy respectivelyBoth studies concluded that the ictal MEG results were inagreementwith invasive andnoninvasive preoperative assess-ments Also the ictal and interictal MEG source localizationswere topographically very similar These findings howeverwere rather preliminary in nature and insufficient informa-tion was available as to assess the reliability of the utility ofMEG in identification of epileptogenic zones

Further support for the role of ictal MEG in presurgicalevaluation is provided by a recent study in children withintractable complex partial epilepsy [96] Over a span of35 years the authors successfully obtained ictal data in8 pediatric patients who subsequently underwent surgicalresection During MEG recordings the patients were either

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

ISRN Radiology 3

Elektra NeuromagTM

3 seconds

600 f

Tcm

Figure 1 The MEG scanner at the Oxford Centre for Human Brain Activity Left Shown is a typical set up where the subject is seatedupright under the scanner A button box is used to record behavioral responses and an eye-tracker supplies additional psychometrical dataMeasurements in supine position are possible and a doctor or nurse can stay within the magnetically shielded room in order to monitor thepatient if needed Middle The brain magnetic fields are recorded with a helmet-shaped array of 102 SQUID devices featuring 3 pick-up coilseach (306 channels in total) Two first-order gradiometers measure two orthogonal spatial gradients of the magnetic field in longitudinal andlatitudinal directions respectively These channels are most sensitive to the tangential neuronal currents in the region below the device Amagnetometer coil (black rectangle) measures the magnetic field and is sensitive to deeper sources Right Typically the cortex-to-detectordistance is 4-5 cm The primary output of the scanner is a trace for each channel representing the magnetic field or gradient as a function oftime (for presentation only 16 channels are shown brain image courtesy of Elekta Neuromag OY Helsinki)

32 Time-Frequency Analysis Viewing brain activity as anoscillatory phenomena there is a long history of studyingfundamental brain rhythms such as theta (4ndash7Hz) or alpha(8ndash13Hz) waves that under certain conditions may indicategeneral states of the brain for example increased alpha isoften seen when subjects close their eyes [32] Robust algo-rithms are now available to analyze neurophysiological sig-nals in terms of oscillations that vary in frequency and timeTypically transient oscillatory brain activity is consideredin the context of well-defined stimuli and is identified aseither an evoked oscillatory response appearing at the samelatency and phase in each trial or an induced oscillatoryresponse appearing with a jitter in phase from one trial toanother and therefore not detectable in the averaged dataBoth types of response are assumed to provide informationabout brain dynamics beyond what can be extracted from theconventional evoked response which includes all frequencies[33 34]

Oscillations at higher frequencies in the beta (15ndash30Hz)and gamma (gt30Hz) bands have received much attention inrecent years Experimental data reveal that synchronizationof activity in these bands is involved in processes such asworking memory semantic processing and sensory-motorintegration and that synchronization may be altered underpathological conditions [35 36] Theoretical considerationssuggest that synchrony of oscillations can facilitate thecoordinated interactions of large neuronal population dis-tributedwithin and across distinct regions of the brain where

interneuron networks are believed to play an important role[37 38]

Of particular relevance to clinical application is event-related desynchronization (ERD) ERD refers to the well-known phenomenon that certain events or stimuli can desyn-chronize or even block parts of the ongoing brain activityThese induced changes can be detected using a techniqueemploying band-pass filtering squaring and averaging inorder to obtain a measure of the oscillatory power in a givenfrequency band of interest ERD is then quantified relativeto the oscillatory power obtained from a reference intervalThe ERD measure may become positive dependent on thenature of task and stimuli in which case one uses the termevent-related synchronization (ERS) ERD (ERS) may beconsidered to be due to a decrease (increase) in synchronyof the underlying neuronal populations and is recognized asa robust marker of normal and abnormal neuronal processesin clinical applications of EEG and MEG [39]

33 Source Estimation Most relevant to clinical MEG issource estimation that is the identification of the brain areasgenerating a given signal recorded by the detectors (alsoknown as magnetic source imaging MSI) Source estimationis ultimately based on the theory of electromagnetism whichis governed by 4 equations known as Maxwellrsquos equationswhich relate the spatial and timederivatives of the electric andmagnetic fields to the charges and currents present within agiven volume defined by its geometry electric permittivity

4 ISRN Radiology

and magnet permeability [40] In the case of electrophysi-ological signals the useful frequency spectrum is typicallybelow 2 kHz and most studies deal with frequency between005Hz and a few hundred hertz As a consequence onecan disregard the time derivatives yielding the quasi-staticapproximation

119861 (119903) sim int119871 (119903 1199031015840

) times 119869 (1199031015840

) 1198891199031015840

(1)

also known as Biot-Savartrsquos law The rule relates the currentflow 119869 at location 1199031015840 within a given source volume to the mag-netic field 119861 observed at position 119903 The integral (continuoussum) runs over the volume containing the currents where thelead field 119871 establishes a link between currents and magneticfield taking into account volume geometry and propertiesTypically one associates the magnetic field with the actualmeasurement in which case the lead field is also dependenton the type position and orientation of the detector usedThen the physical interpretation of the lead field is that itrepresents the sensitivity pattern of a given detector and agiven source volume [21 41]

Conceptually the law constitutes a forward problem thatis knowing the currents and the properties of the embeddingmedium one can calculate the magnetic field in an analyticalor numerical fashion In contrast every source estimationprocedure has to face the so-called inverse problem that isthemathematical transformation that uses the observedmag-netic field as input and determines the sources of the signalsHere there is a problem that is shared with many algo-rithmic proceduresmdashthe inverse problem in MEG is deeplymathematically ambiguous [41 42] No matter how highquality the data is and howmany measurement channels oneuses there can always be silent albeit neurophysiologicallyrelevant currents in the brain that are not recorded by thedetectors The MEG inverse problem is approached prag-matically by simplifying the description of the sources orby adding additional physiologically reasonable assumptionsThe forward model is always needed for source estimationand it is commonly assumed that the current flow in thehead volume can be expressed as the sum of a primary orimpressed current due to the electromotive forces associatedwith biologicalmainly synaptic activity in neural tissue and apassive current flow that results from motion of extracellularcharge carriers [43] Consider

119869 (119903) = 119869impressed (119903) + 119869passive (119903) (2)

331 Equivalent Current Dipoles The arguably simplest anduntil recently most commonly used source estimation tech-nique is the use of equivalent current dipoles (ECD) An ECDcan be viewed as the spatial average of all impressed currentswithin an area of the brain assuming that the arearsquos spatialextent is small compared with the distances to the detectorsEach equivalent dipole is characterized by 6 parametersdescribing the dipolersquos position orientation and strengthwithin the head In general a nonlinear iterative algorithmis applied in order to find the parameters best describingthe data measured In each iteration step the magnetic fields

generated by the dipoles are compared with the fields mea-sured by the detectors and the parameters are adjusted tominimize the field difference quantified by a cost functionTypically a single homogenous sphere is used to approximatethe head volume although other conductormodels have beenused such as ellipsoids several spheres each fitted to thelocal curvature of the head or realistically shaped geometriesbased on individual structural scans

In case of the homogenous sphere the magnetic forwardproblem becomes independent of tissue conductivities andthe spherical symmetries imply that radially oriented currentdipoles become hidden Thus this source model can onlyaccount for tangential dipoles assumed to be primarilylocated in the walls of cortical fissures and sulci but not in theconvexial cortex [21 41 44] ECD approaches vary accordingto the number of dipoles employed (one dipole being themost common choice) the number and type of parametersallowed to change during fit (eg fixed position with rotatingorientation and time-varying strength) and the nature ofsearching or scanning algorithm used to fit parameters todata

332 Imaging Methods Imaging approaches to the inverseproblem explore quasi-continuous estimates of activation bydividing the brain volume into distinct voxels and allocating3 orthogonal equivalent current dipoles to each voxel Severalthousand voxels are usually employed approximating anyarbitrary spatial distribution of post-synaptic currents in thebrain The inverse problem in this case is linear since thesource locations are known leaving only source amplitudes tobe estimated The problem however is severely underdeter-mined given that the number of detectors is small comparedto the number of voxels and regularization is required inorder to restrict the range of allowable source currents whilesimultaneously minimizing the difference betweenmeasuredand predicted magnetic fields

Symbolically

minimum119869

119861 minus 119871 times 119869119901 + 1205821003817100381710038171003817119882model times 119869

1003817100381710038171003817

119901

(3)

where sdot sdot sdot 119901 denotes a norm (roughly strength) of thequantity enclosed In other words one estimates for all voxelsthose current amplitudes that best explain the data and keepa chosen model term also known as a penalty term smallCommonly the voxels are not uniform throughout the cra-nium but restricted to the grey matter or more specifically tothe cerebral cortex In line with the assumption that currentsin the dendritic trunks of pyramidal cells predominantly con-tribute to external fields the dipole orientation may furtherbe constrained to be perpendicular to the local cortical sur-face The parameter 120582 controls the amount of regularizationand its optimal value is usually determined in a data-drivenfashion using an appropriate heuristic approach The weightmatrix 119882model is used to assign additional properties of theinverse solution such as focal sources In essence the weightand the normparameter119901 together define the type of imagingsolution each of which has its merits and disadvantages [44ndash50] The most common choices of norms and weights aresummarized in Table 1

ISRN Radiology 5

Table 1 Comparison of distributed source algorithms often used in clinical MEG research

Name 119901 119882model Comment

MNE 2 1+ simpleminus currents packed towards surface of volume

MNE (weighted) 2 1current-gain + no surface biasminus blurry images

LORETA 2 Laplacian+ + deep source accurateminus estimates superficial sources too deep

MCE 1 1current-gain + focal estimatesminus strong sensitivity to noise

MNE minimum norm estimate (eg [42]) MCE minimum current estimate [46] LORETA low resolution brain electromagnetic tomography [45] +Math-ematical operator representing flux density and gradient flow of a function

333 Beamforming Beamforming approaches to the inverseproblem aim at discriminating between signals originatingfrom a location of interest and those generated elsewhere inthe brain Similar to imaging techniques one assumes thatthe MEG measurement can be modeled by a set of fixedcurrent dipoles distributed throughout the brain In contrastto imaging techniques however one does not attempt to solvea set of underdetermined equations using norm constraintsInstead a beamformer estimates the neural activation at eachvoxel independent of all other locations by means of a spatialfilter Ideally such filter 119865 has complete spatial specificityzeroing the lead field at all locations except the chosen one

119865 (1199030

) times 119871 (119903 1199031015840

) =

1 1199031015840

= 1199030

0 1199031015840

= 1199030

(4)

where 1199030

and 1199031015840 are locations with the source volume and119903 refers to the location of measurement Once the filter hasbeen estimated the neuronal current at position 119903

0

can berecovered from the magnetic fields detected by the sensorarray

119869 (1199030

) = 119865 (1199030

) times 119861 (5)

This expression is commonly interpreted as a virtual electrodethat is one gains knowledge about a current as if an electrodewas attached to the location of interest In practice it isgenerally impossible to have complete attenuation in the stop-band that is locations other than the one of interest and oneaims for designing a filter that is optimal in some sense[44 51]

Arguably the most influential guiding philosophy fordesigning an optimal filter is the linear constrainedminimumvariance approach or LCMV for short [51] The idea is todetermine a filter that minimizes the variance of the sourceat a given location while satisfying the linear response con-straint 119865 times 119871 = 1 ensuring that the signals of interest passby the filter Minimization of source variance which is ameasure of current strength optimally allocates the stop-band response of the filter tominimize the contribution to theoutput due to signals generated by strong sources in the stop-band The mathematical apparatus of beamforming relies onthe assumption that the dipole sources generate mutuallyuncorrelated brain activity Correlated activation cannot bereliably disambiguated from correlated signals at the sensor

level due to volume conduction and field propagation Inother words the sensor readings might be correlated becauseunderlying activations are correlated or the field generated bya source is detected bymore than one sensor at the same timeThe assumption that brain activity is uncorrelated acrossregions might not be physiologically justifiable however ithas been shown that beamformer source estimates are tosome extent robust against correlated sources [44 51 52]

A related approach is known as hardware beamformingexploiting the physical properties of certain detectors typesSensors like planar gradiometers employ two coils measuringorthogonalmagnetic field gradients (spatial variation) at eachsensor location The root-mean-square of the two readingscan be interpreted as the strength of a current dipole locateddirectly below each gradiometer yielding a sensor-level mapof brain activation that can be projected onto the corticalsurface The method is easy to implement and might beapplicable even if the scanner does not have sensors directlysuitable for hardware beamforming [53]

334 Summary Source Estimation The researchers and clin-icians can now draw on awide range of source reconstructionmethods often available in form of commercial and opensoftware packagesThe source images obtained fromdifferentalgorithms are often broadly consistent and may in favor-able circumstances offer meaningful insights into functionalanatomy at the centimeter level However no general sourceimaging method is currently available where the results maydepend on the data and method of analysis In particular thespatial extent of source images may reflect the properties ofthe estimate rather than the true extent of the neural currentStatistical inferences based on source estimates need to bejudged with care as significance does not necessarily implysignificantly nonzero real primary current in a voxel [54] Inany case the interpretation of source estimates should includeconsideration of the nature of the underlying algorithm

4 Artifacts and Artifact Removal

The high sensitivity of MEG implies that the recordings aresensitive to artifacts originating from a variety of physiolog-ical and environmental sources This issue is not trivial andcan become a serious challenge in certain patient populationsdue to amongst others heightened physiological responses

6 ISRN Radiology

lack of control over movements and magnetic materials nec-essary for diagnosis and treatment [4] In general the impactof artifacts may be reduced or controlled to some extentthrough careful preparation and experimental design andexclusion of the most affected data may facilitate analysis andinterpretation However it is impossible to exploit clinicaldata fully without advanced approaches to signal processingthat can reliably separate brain signals and artifacts (Figure 2illustrates the concepts discussed in this section)

41 Physiological Artifacts Arguably the two most dominantartifact sources are the heartbeat and eye blink Althoughthese physiological artifacts can have rather large amplitudessignal separation is challenging as the artificial signals mayresemble activity originating from neuronal sources Regard-ing the eye normal automatic blinking occurs between about5 and 20 times per minute and causes magnetic fields overpredominately frontotemporal regions that can be muchlarger than the magnetic fields of spontaneous and evokedbrain activity The primary artifact is generated by currentflow between the eye and scalp and by excitation of ocularmuscles controlling eye and eyelid movements The artifacttime course recorded by MEG or EEG during a spontaneousblink corresponds to the eyelid motion reaching a closedposition on average 120ms after the blink is initiated andbefore returning to a fully open position approximately240ms after closure In addition to the direct artifact fieldsblink related neuronal activity lasting up to several tenths ofa second is also often observed over posterior parietal andfrontal cortices as well as other areas of the brain associatedwith higher-order processing [57 58] For artifact correctionone generally obtains an independent record of the artifactby measuring the electrooculogram (EOG) with electrodesplaced close to the eyes

Regarding the heart electrical currents moving throughthe organ during a heartbeat give rise to magnetic fields thatare over the chest between two and three orders of magni-tude larger than the brain magnetic fields The time course ofthe cardiac artifact closely corresponds to the propagation ofelectrical activity through the heart muscle Each heartbeatbegins with an impulse from the sinoatrial node the heartrsquosmain pacemaker activating the upper chambers (atria) Thisactivation is seen as the P wave (duration about 80ms) in theelectrocardiogram (ECG) Next the electrical current flowsdown the heart and activates the lower chambers (ventricles)giving rise to a characteristicwave known as theQRS complex(80ndash120ms) Subsequently the electrical current spreadsback over the ventricles in the opposite direction generatinga recovery wave commonly denoted by the letter T (160ms)Typically these currents flowing in the heart are observableto varying degrees in MEG and EEG recordings causingartifacts that may resemble epileptic spikes and slow waves Itis generally assumed that secondary artifacts due to heart beatrelated blood pulsation body movements (ballistocardio-effects) and susceptibility changes are negligible but mayexist in MEG and EEG recordings if the lungs skin or bloodvessels contain magnetic particles [59 page 162] and [60]For artifact correction one generally obtains an independent

record of the artifact by measuring the ECG with electrodesplaced on both arms close to the wrists

Various approaches have been suggested in the past tocorrect for these artifacts often but not exclusively utilizingequivalent current dipoles to model nonneuronal aspectsof the artifacts for example the retinarsquos electrical charge[61] Despite merits however those methods are increas-ingly being replaced by signal projection-based approachesProjection methods are broadly analogous to obtaining a2-dimensional shadow from a 3-dimensional object wherethe direction of illumination determines which features arevisible in the projection Mathematically the data are repre-sented as points (also called vectors) in a space of dimension119899 equal to the number of measurement channels Each 119899-dimensional vector corresponds to one sampled time pointThen the set of new vectors 119881 is calculated to form a basisfor the unwanted signal features that is the artifact to beremoved In other words the set 119881 spans a subspace that isas far as possible occupied by the artifact but not the neuro-physiological signals of interest An artifact-free (clean)measurement is readily obtained through projection

119872119888

= (1 minus 119881 times 119881+

) times 119872 (6)

where 119872 denotes the measurement matrix of dimension119899 times (number of time-points) 1 denotes the 119899-dimensionalidentity matrix and 119881+ is the pseudoinverse of 119881 which hasdimension 119899 times 119896 where 119896 is less than 119873 [62] Dependenton the artifact to be removed 119896 typically assumes a valuebetween 1 and 3

Perhaps the most powerful method to determine theartifact subspace 119881 is a type of blind source separation(BSS) known as independent component analysis (ICA) ICAis entirely data driven and assumes that the observations119872 constitute a linear superposition (mixing) of statisticallyindependent source signals The number of independentsources 119904 is assumed to be the most equal to the number ofmeasurement channels A frequently quoted analogy is theldquococktail party problemrdquo In this scenario party guests holdmultiple simultaneous conversations recorded using micro-phones located at fixed positions around the room Knowingthe number of speakers (s le number of microphones) ICAallows recovering of the individual conversations by unmix-ing the recordings

119878 = 119880 times119872 (7)

where the number of rows of 119878 is equal to the number ofspeakers Each row of 119878 contains the sound from one speakeronly The number of columns of the un-mixing matrix 119880 isequal to the number of microphones and each element of119880 can be interpreted as a weight determining how much anindividual speakerrsquos voice contributed to a given microphone[63 64] Using this analogy the task is then to identify theldquospeakersrdquo who deliver the artifact and guide them out theroom

Over the last 7 years or so several algorithms have beenpublished in order to address this issue in an automatedfashion The main feature that distinguishes the differentapproaches is whether or not an independent EOG andor

ISRN Radiology 7

EOG

CorrectedOriginal

Source power

Original Corrected

Front

Back

RightLeftSi

gnal

(fT

cm) 300

200100

0minus100

minus200

Time (s)4 5 6 7 8

(a)

(b)

(c)

Figure 2 Artifact removal (a) Shown are data from one channel before (black) and after (red) ICA-based correction for eye-blink artifactsidentified by EOG activity (bottom trace) The inset illustrates the 2-dimensional artifact space used for projection (time-courses and spatialtopographies) The first dimension is dominant Note the artifact has similar features as epileptic discharges in that it involves several MEGchannels has sharp peaks and stands out from ongoing background activity (see also [55]) (b) Equivalent current dipole modeling of anevoked somatosensory response in a child before (left) and after (right) SSS-based movement correction After correction the ECD assumesa physiologically plausible location (courtesy of Elekta Neuromag OY Helsinki) (c) Brain activity can be localized accurately despite strongartifacts caused by DBS electrodes (right) Before beamforming-based correction strong nonphysiological interferences outside the brainare observed (left adapted from [56])

ECG recording is needed for artifact identification Table 2shows a brief overview of some of the relevant studies

A recent study in 26 healthy volunteers compared variousartifact detection metrics as well as different flavors of ICAand other blind source separation techniques [70] In gen-eral the authors found automated artifact removal methodspracticable where satisfactory results can be obtained inmostcases irrespective of the detailed nature of the algorithmDependent on type of artifact and data to be cleaned how-ever a combination of approaches employing specific detec-tion metrics might be needed in order to reduce unwantedsignals to an acceptable level It is noteworthy to know thatvalidation has so far been performed in healthy volunteersonly making it difficult to judge when fully automatedmethods will meet clinical standards

42 Head Movement Head or more general subject move-ment during signal acquisition can significantly affect theusability of MEG data due nonneuronal signals arising frommuscle activity artifacts caused by motion related scannervibrations and changes in spatial reference impacting sub-sequent source localization Typically healthy adult patientsare highly motivated and maintain a constant head positionwithin about 2mm standard deviation of the measured headpositions in particular when scans are short Movementrelated artifacts however might increase substantially incertain patient populations and in pediatric measurements[4] Early approaches aimed at preventing head movement asfar as possible with deflatable support cushions bite-bars orother constraints In contrast modern approaches increasingrely on continuous position monitoring in conjunction with

computational methods to correct for head movementsduring data preprocessing

Arguably the most advanced approaches utilize signal-space-separation (SSS) This technique is based on funda-mental properties of quasi-static magnetic fields implyingthat the measurement geometry can be divided into twosource volumes with respect to the sensor array The internalvolume is a sphere whose radius is defined as the distancefrom the head origin to the nearest MEG sensorThe externalvolume is defined as the outside of a spherewhere the radius isthe distance from the head origin to the farthest MEG sensorThen themagnetic fieldmeasured by the sensors (119872) locatedin the current-free space between the internal and externalvolumes can be represented as

119872 = 119872int +119872ext = 119867 times 119860 = [119867int 119867ext] [119860 int119860ext

]

= 119867int times 119860 int + 119867ext times 119860ext

(8)

where 119867 depends on the sensor array and its location withrespect to the two source volumes In contrast the multipolemoments 119860 are device and geometry independent and canbe estimated from the original measurement 119872 Only theinternal terms in the formula above are of interest as they areassumed to represent sources in the brain and a signal-spaceseparated (filtered) signal is obtained by omitting the externalterms thereby suppressing artifacts originating far away fromthe MEG sensors Critically the device independency of 119860allows correction of head movements according to the fol-lowing

119872119904

= 119867119904

times 119860 int (9)

8 ISRN Radiology

Table 2 Overview of some of the relevant ICA-based methods for ocular andor cardiac artifact removal in chronological order

First author Artifact Electrode Sample Comment

Rong [65] Ocular cardiac No 5 One of the first automated removal methods requires a predeterminedartifact template

Okada [66] Ocular Yes 7 Reliable performance for blinks occurring in rapid successionMantini [67] Ocular cardiac No 12 Requires complex parameter tuning can correct environmental artifactsDammers [68] Ocular cardiac Yes 6 Uses signal amplitudes and phases can correct for muscle artifactsKlados [69] Ocular Yes 27 Hybrid regression-ICA approach available as open sourceElectrode indicates whether an electrooculogram andor electrocardiogram measurement is required Sample number of healthy volunteers used for valida-tion

where 119867119904

represents a virtual sensor array locked to thesubjectrsquos head In other words one can calculate the signalsrecorded by a sensor array with respect to where the head isstationary provided that a continuousmovementmonitoringmethod is available [71 72] A temporal extension of thisapproach known as tSSS attempts to identify residual signalscommon to 119872int and 119872ext Such signals typically but notnecessarily represent artifacts originating between the insideand outside volumes and once identified can be removedfrom119872int using a projection as described previously [73]

The SSS approach has been tested extensively wherethe suppression of artifacts originating at a large distance(gt1m) from the scanner matches the theory-based expec-tation [74 75] Note that suppression of nearfields such asocular and cardiac artifacts can be unsatisfactory Excellentresults however have been obtained for tSSS in the case ofspeech artifacts where the movements of tongue lips andjaw and the activity of facial muscles produce magnetic fieldsinterfering with the signals [76] This appears to hold alsofor magnetic fillings and other dental implants significantlyincreasing the number of cases that can be successfullyscanned with MEG

Although movement correction based on SSS has beenvalidated using artificial sources relatively few studies haveinvestigated the effects of head motion in vivo Initiallythe location error of the early somatosensory response toelectrical median nerve stimulation was calculated in a singlesubject making controlled headmovements [77]The authorsreported a 3-fold reduction in localization error comparedto raw data when using tSSS-based movement correction forhead shifts up to 3 cmThe limit of 3 cm is broadly consistentwith a single-case study reporting consistent localization ofepileptic activity when correcting for head movement duringseizure (23 arc length) whereas activity located in differentsublobar regions without correction [78] (Note the MEGresults did not change patientmanagement see Section 51 fora discussion of MEG in epilepsy research)

A recent study systematically investigated the effectsof head motion in a larger sample size [79] The authorsrecorded the neuronal response to auditory and somatosen-sory stimuli in 20 healthy volunteers performing controlledhead movements Moreover magnetized particles wereattached to the subjectrsquos head to simulate nearby interferencesources and a phantom head containing current dipoles wasused for cross validation Confirming and extending previousfindings the results showed that source locations can be

recovered reliably with SSS (error about 5mm) for headshifts not exceeding 3 cm Also using tSSS to remove artifactsdue to nearby sources did not affect the reference dataThe authors pointed out that auditory evoked responsesappeared to be more affected by continuous movement thansomatosensory responses implying that the outcomes ofmotion correction have to be judged in a context dependentmanner

43 Implanted Materials Implanted metal containing ob-jects such as electrodes hydrocephalus shunts or ferromag-netic remnants of neurosurgical operations pose particu-larly high challenges for MEG artifact correction as non-physiological magnetic fields are generated ldquoin siturdquo directlyinterfering with the brain signals The artifacts become evenworse when electrically active devices are involved such asbrain stimulators or pacemakers There is currently no uni-versally applicable method available that corrects for thesetypes of artifacts however promising results are beginningto emerge Two examples may suffice for illustration here

One study used MEG to record epileptic discharges inpatients carrying a vagus nerve stimulator which has provenbeneficial to some epileptic patients considered unsuitablecandidates for resective surgery [80] Using tSSS to clean thedata the authors were able to obtain interpretable signals in 7out of 10 patients The clean signals were subsequently mod-eled using equivalent dipoles in order to localize interictalspikes As a consequence of the MEG results managementchanged to invasive video-EEG (IVEM) in two patientsFurthermore the MEG findings supported the proposedmanagement in four patients demonstrating the feasibility ofMEG in patients with a nerve stimulator

Another study recorded simultaneous MEG and localfield potentials in 17 subjects with Parkinsonrsquos disease (PD)carrying a deep brain stimulator (DBS) to alleviate symptomscaused by degrading motor control [81] Using beamform-ing the authors were able to suppress the high-amplitudeartifacts caused by the DBS wire and electrode and extractinterpretable virtual electrode time-series Subsequent time-frequency analysis pointed to specific synchronization in thegamma band (60ndash90Hz) that stimulates movement throughmodulatory effects in the basal ganglia cortical networkAlthough not of direct clinical relevance this finding is ofimportance for a better understanding of motor control inPD

ISRN Radiology 9

5 Clinical Application Epilepsy

The most well-established clinical application of MEG ispresurgical evaluation in epilepsy Epilepsy is a diverse set ofneurological disorders characterized by the tendency to haverecurring seizures A seizure is the response to an abnormalelectrical discharge in the brain and it describes a varietyof behaviors such as psychic aberrations (eg a sense ofdeja vu) abnormal sensations (eg sensing an intense smell)impairments in speech or speech comprehensions and coor-dinated and involuntary movements Both the nature andseverity of seizure symptoms are dependent on which partsof the brain are affected by the abnormal discharges Seizuresof low severity leave consciousness unimpaired whereas themost severe seizuresmay become life-threatening and requireinstant medical attention [82 83]

Often it is unknown what causes abnormal electricaldischarges however certain risk factors have been identifiedsuch as brain lesions disorders of themetabolic system braininfections and drug and substance abuse amongst others Aseizure typically lasts up to several minutes implying that thetime period that exists between seizures (interictal period)accounts for the majority of a patientrsquos life with epilepsyAlthough most patients may appear symptom free duringinterictal periods abnormal dischargesmay be presentwithinthe brain The anatomic origin of such discharges is oftenalthough not always close to the anatomic origin of seizure-related (ictal) discharges implying a clinical relevance ofinterictal epileptic activity [84]

In industrialized countries epilepsy has an estimatedprevalence of 4ndash10 per 1000 and an incidence of 40ndash70 per100000 and year causing substantial medical expendituresThe disorder is usually controlled but not cured withantiepileptic drugs resulting in seizure freedom in about60 of patients The remaining 40 of affected individualshowever have to be considered pharmacoresistant [85]Surgery is an alternative option in these cases if the seizurescan be traced to abnormal electrical discharges in oneor more clearly delineated brain areas Then the removalor disconnection of the epileptogenic tissues can entail highrates of success Epilepsy surgery however faces substan-tial challenges The ultimate goal is improvement of qual-ity of life while avoiding cognitive or other impairmentscaused by damage to essential functional areas This requiresa substantial multitechnological and possibly invasive pre-operative workup aimed at both the accurate localization ofthe epileptic foci and mapping of nearby essential functionalareas where only the combined results can enable successfulsurgery [86]

51 Localization of Epileptic Foci The potential of MEGto localize epileptogenic regions was communicated to thewider scientific audience over 25 years ago [87] Efforts atconfirming the accuracy of MEG source localizations havebeen addressed from various indirect and direct approachesIndirect validation comes predominantly from studies show-ing colocalization of the magnetic source estimates withknown epileptogenic tissues apparent on structural and func-tional MR imaging or confirmed with invasive intracranial

EEG (IC-EEG) and successful surgical outcomes In contrastthe direct methods reflect studies utilizing simultaneousMEG and IC-EEG recordings or simulated epileptic dis-charges generated by implanted dipoles [85 88] Similar toother non-invasive approaches to preoperative evaluation aconsiderable amount of the MEG work has been based oninterictal discharges (an example is shown in Figure 3(a)[55 89ndash91]) However the importance to determine thelocation of the onset of ictal activity associated with clinicalseizures has prompted researcher to address methodologicalchallenges and validate ictal MEG [92]

In one of the first larger studies of its kind successfulictal MEG recordings were made in 6 out of 20 patientswith intractable complex partial seizures [94] All patientshad neocortical epilepsies and data collection took placeover a span of 3 years In order to capture ictal-onsetactivity a physician remained in the magnetically shieldedroom with each patient and retracted the probe soon after aseizure occurred allowing the patient to move freely Duringrecording a secondphysicianmonitored real-time displays ofthe MEG and EEG waveforms and triggered data collectionwhen abnormal discharges were detected On average theictal scans took several hours and the patients were allowedlight sleep unless it was necessary to reposition the head or ifsleep to wake transitions were felt necessary to obtain ictaldata This allowed the authors to record a few seconds ofictal data although their procedure had a limited capabilityto track seizure spread

The authors used single equivalent current dipole tolocalize sources underlying the MEG field patterns detectedduring ictal and interictal data segments Compared to IC-EEG the authors reported that ictalMEGprovided localizinginformation that was superior to interictal MEG in three ofthe six patients Localization of ictal onset byMEGwas foundto be equivalent to invasive EEG in five of the six patientsIn one patient ictal-intracranial EEG showed nonlocalizingdiffuse seizure related activity whereas the MEG resultspointed to a circumscribed ictal onset zone in right prefrontalcortices A resection was performed based in part on theMEG findings leaving the patient seizure free and withoutpostoperative neurocognitive deficit over 2 years of followupDespite a small number of ictal recordings the authors rightlyconcluded that ictal MEG might be superior to invasiverecordings under certain conditions

The study furthered two previous investigations in 4patients [95] and 3 patients [92] with epilepsy respectivelyBoth studies concluded that the ictal MEG results were inagreementwith invasive andnoninvasive preoperative assess-ments Also the ictal and interictal MEG source localizationswere topographically very similar These findings howeverwere rather preliminary in nature and insufficient informa-tion was available as to assess the reliability of the utility ofMEG in identification of epileptogenic zones

Further support for the role of ictal MEG in presurgicalevaluation is provided by a recent study in children withintractable complex partial epilepsy [96] Over a span of35 years the authors successfully obtained ictal data in8 pediatric patients who subsequently underwent surgicalresection During MEG recordings the patients were either

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

4 ISRN Radiology

and magnet permeability [40] In the case of electrophysi-ological signals the useful frequency spectrum is typicallybelow 2 kHz and most studies deal with frequency between005Hz and a few hundred hertz As a consequence onecan disregard the time derivatives yielding the quasi-staticapproximation

119861 (119903) sim int119871 (119903 1199031015840

) times 119869 (1199031015840

) 1198891199031015840

(1)

also known as Biot-Savartrsquos law The rule relates the currentflow 119869 at location 1199031015840 within a given source volume to the mag-netic field 119861 observed at position 119903 The integral (continuoussum) runs over the volume containing the currents where thelead field 119871 establishes a link between currents and magneticfield taking into account volume geometry and propertiesTypically one associates the magnetic field with the actualmeasurement in which case the lead field is also dependenton the type position and orientation of the detector usedThen the physical interpretation of the lead field is that itrepresents the sensitivity pattern of a given detector and agiven source volume [21 41]

Conceptually the law constitutes a forward problem thatis knowing the currents and the properties of the embeddingmedium one can calculate the magnetic field in an analyticalor numerical fashion In contrast every source estimationprocedure has to face the so-called inverse problem that isthemathematical transformation that uses the observedmag-netic field as input and determines the sources of the signalsHere there is a problem that is shared with many algo-rithmic proceduresmdashthe inverse problem in MEG is deeplymathematically ambiguous [41 42] No matter how highquality the data is and howmany measurement channels oneuses there can always be silent albeit neurophysiologicallyrelevant currents in the brain that are not recorded by thedetectors The MEG inverse problem is approached prag-matically by simplifying the description of the sources orby adding additional physiologically reasonable assumptionsThe forward model is always needed for source estimationand it is commonly assumed that the current flow in thehead volume can be expressed as the sum of a primary orimpressed current due to the electromotive forces associatedwith biologicalmainly synaptic activity in neural tissue and apassive current flow that results from motion of extracellularcharge carriers [43] Consider

119869 (119903) = 119869impressed (119903) + 119869passive (119903) (2)

331 Equivalent Current Dipoles The arguably simplest anduntil recently most commonly used source estimation tech-nique is the use of equivalent current dipoles (ECD) An ECDcan be viewed as the spatial average of all impressed currentswithin an area of the brain assuming that the arearsquos spatialextent is small compared with the distances to the detectorsEach equivalent dipole is characterized by 6 parametersdescribing the dipolersquos position orientation and strengthwithin the head In general a nonlinear iterative algorithmis applied in order to find the parameters best describingthe data measured In each iteration step the magnetic fields

generated by the dipoles are compared with the fields mea-sured by the detectors and the parameters are adjusted tominimize the field difference quantified by a cost functionTypically a single homogenous sphere is used to approximatethe head volume although other conductormodels have beenused such as ellipsoids several spheres each fitted to thelocal curvature of the head or realistically shaped geometriesbased on individual structural scans

In case of the homogenous sphere the magnetic forwardproblem becomes independent of tissue conductivities andthe spherical symmetries imply that radially oriented currentdipoles become hidden Thus this source model can onlyaccount for tangential dipoles assumed to be primarilylocated in the walls of cortical fissures and sulci but not in theconvexial cortex [21 41 44] ECD approaches vary accordingto the number of dipoles employed (one dipole being themost common choice) the number and type of parametersallowed to change during fit (eg fixed position with rotatingorientation and time-varying strength) and the nature ofsearching or scanning algorithm used to fit parameters todata

332 Imaging Methods Imaging approaches to the inverseproblem explore quasi-continuous estimates of activation bydividing the brain volume into distinct voxels and allocating3 orthogonal equivalent current dipoles to each voxel Severalthousand voxels are usually employed approximating anyarbitrary spatial distribution of post-synaptic currents in thebrain The inverse problem in this case is linear since thesource locations are known leaving only source amplitudes tobe estimated The problem however is severely underdeter-mined given that the number of detectors is small comparedto the number of voxels and regularization is required inorder to restrict the range of allowable source currents whilesimultaneously minimizing the difference betweenmeasuredand predicted magnetic fields

Symbolically

minimum119869

119861 minus 119871 times 119869119901 + 1205821003817100381710038171003817119882model times 119869

1003817100381710038171003817

119901

(3)

where sdot sdot sdot 119901 denotes a norm (roughly strength) of thequantity enclosed In other words one estimates for all voxelsthose current amplitudes that best explain the data and keepa chosen model term also known as a penalty term smallCommonly the voxels are not uniform throughout the cra-nium but restricted to the grey matter or more specifically tothe cerebral cortex In line with the assumption that currentsin the dendritic trunks of pyramidal cells predominantly con-tribute to external fields the dipole orientation may furtherbe constrained to be perpendicular to the local cortical sur-face The parameter 120582 controls the amount of regularizationand its optimal value is usually determined in a data-drivenfashion using an appropriate heuristic approach The weightmatrix 119882model is used to assign additional properties of theinverse solution such as focal sources In essence the weightand the normparameter119901 together define the type of imagingsolution each of which has its merits and disadvantages [44ndash50] The most common choices of norms and weights aresummarized in Table 1

ISRN Radiology 5

Table 1 Comparison of distributed source algorithms often used in clinical MEG research

Name 119901 119882model Comment

MNE 2 1+ simpleminus currents packed towards surface of volume

MNE (weighted) 2 1current-gain + no surface biasminus blurry images

LORETA 2 Laplacian+ + deep source accurateminus estimates superficial sources too deep

MCE 1 1current-gain + focal estimatesminus strong sensitivity to noise

MNE minimum norm estimate (eg [42]) MCE minimum current estimate [46] LORETA low resolution brain electromagnetic tomography [45] +Math-ematical operator representing flux density and gradient flow of a function

333 Beamforming Beamforming approaches to the inverseproblem aim at discriminating between signals originatingfrom a location of interest and those generated elsewhere inthe brain Similar to imaging techniques one assumes thatthe MEG measurement can be modeled by a set of fixedcurrent dipoles distributed throughout the brain In contrastto imaging techniques however one does not attempt to solvea set of underdetermined equations using norm constraintsInstead a beamformer estimates the neural activation at eachvoxel independent of all other locations by means of a spatialfilter Ideally such filter 119865 has complete spatial specificityzeroing the lead field at all locations except the chosen one

119865 (1199030

) times 119871 (119903 1199031015840

) =

1 1199031015840

= 1199030

0 1199031015840

= 1199030

(4)

where 1199030

and 1199031015840 are locations with the source volume and119903 refers to the location of measurement Once the filter hasbeen estimated the neuronal current at position 119903

0

can berecovered from the magnetic fields detected by the sensorarray

119869 (1199030

) = 119865 (1199030

) times 119861 (5)

This expression is commonly interpreted as a virtual electrodethat is one gains knowledge about a current as if an electrodewas attached to the location of interest In practice it isgenerally impossible to have complete attenuation in the stop-band that is locations other than the one of interest and oneaims for designing a filter that is optimal in some sense[44 51]

Arguably the most influential guiding philosophy fordesigning an optimal filter is the linear constrainedminimumvariance approach or LCMV for short [51] The idea is todetermine a filter that minimizes the variance of the sourceat a given location while satisfying the linear response con-straint 119865 times 119871 = 1 ensuring that the signals of interest passby the filter Minimization of source variance which is ameasure of current strength optimally allocates the stop-band response of the filter tominimize the contribution to theoutput due to signals generated by strong sources in the stop-band The mathematical apparatus of beamforming relies onthe assumption that the dipole sources generate mutuallyuncorrelated brain activity Correlated activation cannot bereliably disambiguated from correlated signals at the sensor

level due to volume conduction and field propagation Inother words the sensor readings might be correlated becauseunderlying activations are correlated or the field generated bya source is detected bymore than one sensor at the same timeThe assumption that brain activity is uncorrelated acrossregions might not be physiologically justifiable however ithas been shown that beamformer source estimates are tosome extent robust against correlated sources [44 51 52]

A related approach is known as hardware beamformingexploiting the physical properties of certain detectors typesSensors like planar gradiometers employ two coils measuringorthogonalmagnetic field gradients (spatial variation) at eachsensor location The root-mean-square of the two readingscan be interpreted as the strength of a current dipole locateddirectly below each gradiometer yielding a sensor-level mapof brain activation that can be projected onto the corticalsurface The method is easy to implement and might beapplicable even if the scanner does not have sensors directlysuitable for hardware beamforming [53]

334 Summary Source Estimation The researchers and clin-icians can now draw on awide range of source reconstructionmethods often available in form of commercial and opensoftware packagesThe source images obtained fromdifferentalgorithms are often broadly consistent and may in favor-able circumstances offer meaningful insights into functionalanatomy at the centimeter level However no general sourceimaging method is currently available where the results maydepend on the data and method of analysis In particular thespatial extent of source images may reflect the properties ofthe estimate rather than the true extent of the neural currentStatistical inferences based on source estimates need to bejudged with care as significance does not necessarily implysignificantly nonzero real primary current in a voxel [54] Inany case the interpretation of source estimates should includeconsideration of the nature of the underlying algorithm

4 Artifacts and Artifact Removal

The high sensitivity of MEG implies that the recordings aresensitive to artifacts originating from a variety of physiolog-ical and environmental sources This issue is not trivial andcan become a serious challenge in certain patient populationsdue to amongst others heightened physiological responses

6 ISRN Radiology

lack of control over movements and magnetic materials nec-essary for diagnosis and treatment [4] In general the impactof artifacts may be reduced or controlled to some extentthrough careful preparation and experimental design andexclusion of the most affected data may facilitate analysis andinterpretation However it is impossible to exploit clinicaldata fully without advanced approaches to signal processingthat can reliably separate brain signals and artifacts (Figure 2illustrates the concepts discussed in this section)

41 Physiological Artifacts Arguably the two most dominantartifact sources are the heartbeat and eye blink Althoughthese physiological artifacts can have rather large amplitudessignal separation is challenging as the artificial signals mayresemble activity originating from neuronal sources Regard-ing the eye normal automatic blinking occurs between about5 and 20 times per minute and causes magnetic fields overpredominately frontotemporal regions that can be muchlarger than the magnetic fields of spontaneous and evokedbrain activity The primary artifact is generated by currentflow between the eye and scalp and by excitation of ocularmuscles controlling eye and eyelid movements The artifacttime course recorded by MEG or EEG during a spontaneousblink corresponds to the eyelid motion reaching a closedposition on average 120ms after the blink is initiated andbefore returning to a fully open position approximately240ms after closure In addition to the direct artifact fieldsblink related neuronal activity lasting up to several tenths ofa second is also often observed over posterior parietal andfrontal cortices as well as other areas of the brain associatedwith higher-order processing [57 58] For artifact correctionone generally obtains an independent record of the artifactby measuring the electrooculogram (EOG) with electrodesplaced close to the eyes

Regarding the heart electrical currents moving throughthe organ during a heartbeat give rise to magnetic fields thatare over the chest between two and three orders of magni-tude larger than the brain magnetic fields The time course ofthe cardiac artifact closely corresponds to the propagation ofelectrical activity through the heart muscle Each heartbeatbegins with an impulse from the sinoatrial node the heartrsquosmain pacemaker activating the upper chambers (atria) Thisactivation is seen as the P wave (duration about 80ms) in theelectrocardiogram (ECG) Next the electrical current flowsdown the heart and activates the lower chambers (ventricles)giving rise to a characteristicwave known as theQRS complex(80ndash120ms) Subsequently the electrical current spreadsback over the ventricles in the opposite direction generatinga recovery wave commonly denoted by the letter T (160ms)Typically these currents flowing in the heart are observableto varying degrees in MEG and EEG recordings causingartifacts that may resemble epileptic spikes and slow waves Itis generally assumed that secondary artifacts due to heart beatrelated blood pulsation body movements (ballistocardio-effects) and susceptibility changes are negligible but mayexist in MEG and EEG recordings if the lungs skin or bloodvessels contain magnetic particles [59 page 162] and [60]For artifact correction one generally obtains an independent

record of the artifact by measuring the ECG with electrodesplaced on both arms close to the wrists

Various approaches have been suggested in the past tocorrect for these artifacts often but not exclusively utilizingequivalent current dipoles to model nonneuronal aspectsof the artifacts for example the retinarsquos electrical charge[61] Despite merits however those methods are increas-ingly being replaced by signal projection-based approachesProjection methods are broadly analogous to obtaining a2-dimensional shadow from a 3-dimensional object wherethe direction of illumination determines which features arevisible in the projection Mathematically the data are repre-sented as points (also called vectors) in a space of dimension119899 equal to the number of measurement channels Each 119899-dimensional vector corresponds to one sampled time pointThen the set of new vectors 119881 is calculated to form a basisfor the unwanted signal features that is the artifact to beremoved In other words the set 119881 spans a subspace that isas far as possible occupied by the artifact but not the neuro-physiological signals of interest An artifact-free (clean)measurement is readily obtained through projection

119872119888

= (1 minus 119881 times 119881+

) times 119872 (6)

where 119872 denotes the measurement matrix of dimension119899 times (number of time-points) 1 denotes the 119899-dimensionalidentity matrix and 119881+ is the pseudoinverse of 119881 which hasdimension 119899 times 119896 where 119896 is less than 119873 [62] Dependenton the artifact to be removed 119896 typically assumes a valuebetween 1 and 3

Perhaps the most powerful method to determine theartifact subspace 119881 is a type of blind source separation(BSS) known as independent component analysis (ICA) ICAis entirely data driven and assumes that the observations119872 constitute a linear superposition (mixing) of statisticallyindependent source signals The number of independentsources 119904 is assumed to be the most equal to the number ofmeasurement channels A frequently quoted analogy is theldquococktail party problemrdquo In this scenario party guests holdmultiple simultaneous conversations recorded using micro-phones located at fixed positions around the room Knowingthe number of speakers (s le number of microphones) ICAallows recovering of the individual conversations by unmix-ing the recordings

119878 = 119880 times119872 (7)

where the number of rows of 119878 is equal to the number ofspeakers Each row of 119878 contains the sound from one speakeronly The number of columns of the un-mixing matrix 119880 isequal to the number of microphones and each element of119880 can be interpreted as a weight determining how much anindividual speakerrsquos voice contributed to a given microphone[63 64] Using this analogy the task is then to identify theldquospeakersrdquo who deliver the artifact and guide them out theroom

Over the last 7 years or so several algorithms have beenpublished in order to address this issue in an automatedfashion The main feature that distinguishes the differentapproaches is whether or not an independent EOG andor

ISRN Radiology 7

EOG

CorrectedOriginal

Source power

Original Corrected

Front

Back

RightLeftSi

gnal

(fT

cm) 300

200100

0minus100

minus200

Time (s)4 5 6 7 8

(a)

(b)

(c)

Figure 2 Artifact removal (a) Shown are data from one channel before (black) and after (red) ICA-based correction for eye-blink artifactsidentified by EOG activity (bottom trace) The inset illustrates the 2-dimensional artifact space used for projection (time-courses and spatialtopographies) The first dimension is dominant Note the artifact has similar features as epileptic discharges in that it involves several MEGchannels has sharp peaks and stands out from ongoing background activity (see also [55]) (b) Equivalent current dipole modeling of anevoked somatosensory response in a child before (left) and after (right) SSS-based movement correction After correction the ECD assumesa physiologically plausible location (courtesy of Elekta Neuromag OY Helsinki) (c) Brain activity can be localized accurately despite strongartifacts caused by DBS electrodes (right) Before beamforming-based correction strong nonphysiological interferences outside the brainare observed (left adapted from [56])

ECG recording is needed for artifact identification Table 2shows a brief overview of some of the relevant studies

A recent study in 26 healthy volunteers compared variousartifact detection metrics as well as different flavors of ICAand other blind source separation techniques [70] In gen-eral the authors found automated artifact removal methodspracticable where satisfactory results can be obtained inmostcases irrespective of the detailed nature of the algorithmDependent on type of artifact and data to be cleaned how-ever a combination of approaches employing specific detec-tion metrics might be needed in order to reduce unwantedsignals to an acceptable level It is noteworthy to know thatvalidation has so far been performed in healthy volunteersonly making it difficult to judge when fully automatedmethods will meet clinical standards

42 Head Movement Head or more general subject move-ment during signal acquisition can significantly affect theusability of MEG data due nonneuronal signals arising frommuscle activity artifacts caused by motion related scannervibrations and changes in spatial reference impacting sub-sequent source localization Typically healthy adult patientsare highly motivated and maintain a constant head positionwithin about 2mm standard deviation of the measured headpositions in particular when scans are short Movementrelated artifacts however might increase substantially incertain patient populations and in pediatric measurements[4] Early approaches aimed at preventing head movement asfar as possible with deflatable support cushions bite-bars orother constraints In contrast modern approaches increasingrely on continuous position monitoring in conjunction with

computational methods to correct for head movementsduring data preprocessing

Arguably the most advanced approaches utilize signal-space-separation (SSS) This technique is based on funda-mental properties of quasi-static magnetic fields implyingthat the measurement geometry can be divided into twosource volumes with respect to the sensor array The internalvolume is a sphere whose radius is defined as the distancefrom the head origin to the nearest MEG sensorThe externalvolume is defined as the outside of a spherewhere the radius isthe distance from the head origin to the farthest MEG sensorThen themagnetic fieldmeasured by the sensors (119872) locatedin the current-free space between the internal and externalvolumes can be represented as

119872 = 119872int +119872ext = 119867 times 119860 = [119867int 119867ext] [119860 int119860ext

]

= 119867int times 119860 int + 119867ext times 119860ext

(8)

where 119867 depends on the sensor array and its location withrespect to the two source volumes In contrast the multipolemoments 119860 are device and geometry independent and canbe estimated from the original measurement 119872 Only theinternal terms in the formula above are of interest as they areassumed to represent sources in the brain and a signal-spaceseparated (filtered) signal is obtained by omitting the externalterms thereby suppressing artifacts originating far away fromthe MEG sensors Critically the device independency of 119860allows correction of head movements according to the fol-lowing

119872119904

= 119867119904

times 119860 int (9)

8 ISRN Radiology

Table 2 Overview of some of the relevant ICA-based methods for ocular andor cardiac artifact removal in chronological order

First author Artifact Electrode Sample Comment

Rong [65] Ocular cardiac No 5 One of the first automated removal methods requires a predeterminedartifact template

Okada [66] Ocular Yes 7 Reliable performance for blinks occurring in rapid successionMantini [67] Ocular cardiac No 12 Requires complex parameter tuning can correct environmental artifactsDammers [68] Ocular cardiac Yes 6 Uses signal amplitudes and phases can correct for muscle artifactsKlados [69] Ocular Yes 27 Hybrid regression-ICA approach available as open sourceElectrode indicates whether an electrooculogram andor electrocardiogram measurement is required Sample number of healthy volunteers used for valida-tion

where 119867119904

represents a virtual sensor array locked to thesubjectrsquos head In other words one can calculate the signalsrecorded by a sensor array with respect to where the head isstationary provided that a continuousmovementmonitoringmethod is available [71 72] A temporal extension of thisapproach known as tSSS attempts to identify residual signalscommon to 119872int and 119872ext Such signals typically but notnecessarily represent artifacts originating between the insideand outside volumes and once identified can be removedfrom119872int using a projection as described previously [73]

The SSS approach has been tested extensively wherethe suppression of artifacts originating at a large distance(gt1m) from the scanner matches the theory-based expec-tation [74 75] Note that suppression of nearfields such asocular and cardiac artifacts can be unsatisfactory Excellentresults however have been obtained for tSSS in the case ofspeech artifacts where the movements of tongue lips andjaw and the activity of facial muscles produce magnetic fieldsinterfering with the signals [76] This appears to hold alsofor magnetic fillings and other dental implants significantlyincreasing the number of cases that can be successfullyscanned with MEG

Although movement correction based on SSS has beenvalidated using artificial sources relatively few studies haveinvestigated the effects of head motion in vivo Initiallythe location error of the early somatosensory response toelectrical median nerve stimulation was calculated in a singlesubject making controlled headmovements [77]The authorsreported a 3-fold reduction in localization error comparedto raw data when using tSSS-based movement correction forhead shifts up to 3 cmThe limit of 3 cm is broadly consistentwith a single-case study reporting consistent localization ofepileptic activity when correcting for head movement duringseizure (23 arc length) whereas activity located in differentsublobar regions without correction [78] (Note the MEGresults did not change patientmanagement see Section 51 fora discussion of MEG in epilepsy research)

A recent study systematically investigated the effectsof head motion in a larger sample size [79] The authorsrecorded the neuronal response to auditory and somatosen-sory stimuli in 20 healthy volunteers performing controlledhead movements Moreover magnetized particles wereattached to the subjectrsquos head to simulate nearby interferencesources and a phantom head containing current dipoles wasused for cross validation Confirming and extending previousfindings the results showed that source locations can be

recovered reliably with SSS (error about 5mm) for headshifts not exceeding 3 cm Also using tSSS to remove artifactsdue to nearby sources did not affect the reference dataThe authors pointed out that auditory evoked responsesappeared to be more affected by continuous movement thansomatosensory responses implying that the outcomes ofmotion correction have to be judged in a context dependentmanner

43 Implanted Materials Implanted metal containing ob-jects such as electrodes hydrocephalus shunts or ferromag-netic remnants of neurosurgical operations pose particu-larly high challenges for MEG artifact correction as non-physiological magnetic fields are generated ldquoin siturdquo directlyinterfering with the brain signals The artifacts become evenworse when electrically active devices are involved such asbrain stimulators or pacemakers There is currently no uni-versally applicable method available that corrects for thesetypes of artifacts however promising results are beginningto emerge Two examples may suffice for illustration here

One study used MEG to record epileptic discharges inpatients carrying a vagus nerve stimulator which has provenbeneficial to some epileptic patients considered unsuitablecandidates for resective surgery [80] Using tSSS to clean thedata the authors were able to obtain interpretable signals in 7out of 10 patients The clean signals were subsequently mod-eled using equivalent dipoles in order to localize interictalspikes As a consequence of the MEG results managementchanged to invasive video-EEG (IVEM) in two patientsFurthermore the MEG findings supported the proposedmanagement in four patients demonstrating the feasibility ofMEG in patients with a nerve stimulator

Another study recorded simultaneous MEG and localfield potentials in 17 subjects with Parkinsonrsquos disease (PD)carrying a deep brain stimulator (DBS) to alleviate symptomscaused by degrading motor control [81] Using beamform-ing the authors were able to suppress the high-amplitudeartifacts caused by the DBS wire and electrode and extractinterpretable virtual electrode time-series Subsequent time-frequency analysis pointed to specific synchronization in thegamma band (60ndash90Hz) that stimulates movement throughmodulatory effects in the basal ganglia cortical networkAlthough not of direct clinical relevance this finding is ofimportance for a better understanding of motor control inPD

ISRN Radiology 9

5 Clinical Application Epilepsy

The most well-established clinical application of MEG ispresurgical evaluation in epilepsy Epilepsy is a diverse set ofneurological disorders characterized by the tendency to haverecurring seizures A seizure is the response to an abnormalelectrical discharge in the brain and it describes a varietyof behaviors such as psychic aberrations (eg a sense ofdeja vu) abnormal sensations (eg sensing an intense smell)impairments in speech or speech comprehensions and coor-dinated and involuntary movements Both the nature andseverity of seizure symptoms are dependent on which partsof the brain are affected by the abnormal discharges Seizuresof low severity leave consciousness unimpaired whereas themost severe seizuresmay become life-threatening and requireinstant medical attention [82 83]

Often it is unknown what causes abnormal electricaldischarges however certain risk factors have been identifiedsuch as brain lesions disorders of themetabolic system braininfections and drug and substance abuse amongst others Aseizure typically lasts up to several minutes implying that thetime period that exists between seizures (interictal period)accounts for the majority of a patientrsquos life with epilepsyAlthough most patients may appear symptom free duringinterictal periods abnormal dischargesmay be presentwithinthe brain The anatomic origin of such discharges is oftenalthough not always close to the anatomic origin of seizure-related (ictal) discharges implying a clinical relevance ofinterictal epileptic activity [84]

In industrialized countries epilepsy has an estimatedprevalence of 4ndash10 per 1000 and an incidence of 40ndash70 per100000 and year causing substantial medical expendituresThe disorder is usually controlled but not cured withantiepileptic drugs resulting in seizure freedom in about60 of patients The remaining 40 of affected individualshowever have to be considered pharmacoresistant [85]Surgery is an alternative option in these cases if the seizurescan be traced to abnormal electrical discharges in oneor more clearly delineated brain areas Then the removalor disconnection of the epileptogenic tissues can entail highrates of success Epilepsy surgery however faces substan-tial challenges The ultimate goal is improvement of qual-ity of life while avoiding cognitive or other impairmentscaused by damage to essential functional areas This requiresa substantial multitechnological and possibly invasive pre-operative workup aimed at both the accurate localization ofthe epileptic foci and mapping of nearby essential functionalareas where only the combined results can enable successfulsurgery [86]

51 Localization of Epileptic Foci The potential of MEGto localize epileptogenic regions was communicated to thewider scientific audience over 25 years ago [87] Efforts atconfirming the accuracy of MEG source localizations havebeen addressed from various indirect and direct approachesIndirect validation comes predominantly from studies show-ing colocalization of the magnetic source estimates withknown epileptogenic tissues apparent on structural and func-tional MR imaging or confirmed with invasive intracranial

EEG (IC-EEG) and successful surgical outcomes In contrastthe direct methods reflect studies utilizing simultaneousMEG and IC-EEG recordings or simulated epileptic dis-charges generated by implanted dipoles [85 88] Similar toother non-invasive approaches to preoperative evaluation aconsiderable amount of the MEG work has been based oninterictal discharges (an example is shown in Figure 3(a)[55 89ndash91]) However the importance to determine thelocation of the onset of ictal activity associated with clinicalseizures has prompted researcher to address methodologicalchallenges and validate ictal MEG [92]

In one of the first larger studies of its kind successfulictal MEG recordings were made in 6 out of 20 patientswith intractable complex partial seizures [94] All patientshad neocortical epilepsies and data collection took placeover a span of 3 years In order to capture ictal-onsetactivity a physician remained in the magnetically shieldedroom with each patient and retracted the probe soon after aseizure occurred allowing the patient to move freely Duringrecording a secondphysicianmonitored real-time displays ofthe MEG and EEG waveforms and triggered data collectionwhen abnormal discharges were detected On average theictal scans took several hours and the patients were allowedlight sleep unless it was necessary to reposition the head or ifsleep to wake transitions were felt necessary to obtain ictaldata This allowed the authors to record a few seconds ofictal data although their procedure had a limited capabilityto track seizure spread

The authors used single equivalent current dipole tolocalize sources underlying the MEG field patterns detectedduring ictal and interictal data segments Compared to IC-EEG the authors reported that ictalMEGprovided localizinginformation that was superior to interictal MEG in three ofthe six patients Localization of ictal onset byMEGwas foundto be equivalent to invasive EEG in five of the six patientsIn one patient ictal-intracranial EEG showed nonlocalizingdiffuse seizure related activity whereas the MEG resultspointed to a circumscribed ictal onset zone in right prefrontalcortices A resection was performed based in part on theMEG findings leaving the patient seizure free and withoutpostoperative neurocognitive deficit over 2 years of followupDespite a small number of ictal recordings the authors rightlyconcluded that ictal MEG might be superior to invasiverecordings under certain conditions

The study furthered two previous investigations in 4patients [95] and 3 patients [92] with epilepsy respectivelyBoth studies concluded that the ictal MEG results were inagreementwith invasive andnoninvasive preoperative assess-ments Also the ictal and interictal MEG source localizationswere topographically very similar These findings howeverwere rather preliminary in nature and insufficient informa-tion was available as to assess the reliability of the utility ofMEG in identification of epileptogenic zones

Further support for the role of ictal MEG in presurgicalevaluation is provided by a recent study in children withintractable complex partial epilepsy [96] Over a span of35 years the authors successfully obtained ictal data in8 pediatric patients who subsequently underwent surgicalresection During MEG recordings the patients were either

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

ISRN Radiology 5

Table 1 Comparison of distributed source algorithms often used in clinical MEG research

Name 119901 119882model Comment

MNE 2 1+ simpleminus currents packed towards surface of volume

MNE (weighted) 2 1current-gain + no surface biasminus blurry images

LORETA 2 Laplacian+ + deep source accurateminus estimates superficial sources too deep

MCE 1 1current-gain + focal estimatesminus strong sensitivity to noise

MNE minimum norm estimate (eg [42]) MCE minimum current estimate [46] LORETA low resolution brain electromagnetic tomography [45] +Math-ematical operator representing flux density and gradient flow of a function

333 Beamforming Beamforming approaches to the inverseproblem aim at discriminating between signals originatingfrom a location of interest and those generated elsewhere inthe brain Similar to imaging techniques one assumes thatthe MEG measurement can be modeled by a set of fixedcurrent dipoles distributed throughout the brain In contrastto imaging techniques however one does not attempt to solvea set of underdetermined equations using norm constraintsInstead a beamformer estimates the neural activation at eachvoxel independent of all other locations by means of a spatialfilter Ideally such filter 119865 has complete spatial specificityzeroing the lead field at all locations except the chosen one

119865 (1199030

) times 119871 (119903 1199031015840

) =

1 1199031015840

= 1199030

0 1199031015840

= 1199030

(4)

where 1199030

and 1199031015840 are locations with the source volume and119903 refers to the location of measurement Once the filter hasbeen estimated the neuronal current at position 119903

0

can berecovered from the magnetic fields detected by the sensorarray

119869 (1199030

) = 119865 (1199030

) times 119861 (5)

This expression is commonly interpreted as a virtual electrodethat is one gains knowledge about a current as if an electrodewas attached to the location of interest In practice it isgenerally impossible to have complete attenuation in the stop-band that is locations other than the one of interest and oneaims for designing a filter that is optimal in some sense[44 51]

Arguably the most influential guiding philosophy fordesigning an optimal filter is the linear constrainedminimumvariance approach or LCMV for short [51] The idea is todetermine a filter that minimizes the variance of the sourceat a given location while satisfying the linear response con-straint 119865 times 119871 = 1 ensuring that the signals of interest passby the filter Minimization of source variance which is ameasure of current strength optimally allocates the stop-band response of the filter tominimize the contribution to theoutput due to signals generated by strong sources in the stop-band The mathematical apparatus of beamforming relies onthe assumption that the dipole sources generate mutuallyuncorrelated brain activity Correlated activation cannot bereliably disambiguated from correlated signals at the sensor

level due to volume conduction and field propagation Inother words the sensor readings might be correlated becauseunderlying activations are correlated or the field generated bya source is detected bymore than one sensor at the same timeThe assumption that brain activity is uncorrelated acrossregions might not be physiologically justifiable however ithas been shown that beamformer source estimates are tosome extent robust against correlated sources [44 51 52]

A related approach is known as hardware beamformingexploiting the physical properties of certain detectors typesSensors like planar gradiometers employ two coils measuringorthogonalmagnetic field gradients (spatial variation) at eachsensor location The root-mean-square of the two readingscan be interpreted as the strength of a current dipole locateddirectly below each gradiometer yielding a sensor-level mapof brain activation that can be projected onto the corticalsurface The method is easy to implement and might beapplicable even if the scanner does not have sensors directlysuitable for hardware beamforming [53]

334 Summary Source Estimation The researchers and clin-icians can now draw on awide range of source reconstructionmethods often available in form of commercial and opensoftware packagesThe source images obtained fromdifferentalgorithms are often broadly consistent and may in favor-able circumstances offer meaningful insights into functionalanatomy at the centimeter level However no general sourceimaging method is currently available where the results maydepend on the data and method of analysis In particular thespatial extent of source images may reflect the properties ofthe estimate rather than the true extent of the neural currentStatistical inferences based on source estimates need to bejudged with care as significance does not necessarily implysignificantly nonzero real primary current in a voxel [54] Inany case the interpretation of source estimates should includeconsideration of the nature of the underlying algorithm

4 Artifacts and Artifact Removal

The high sensitivity of MEG implies that the recordings aresensitive to artifacts originating from a variety of physiolog-ical and environmental sources This issue is not trivial andcan become a serious challenge in certain patient populationsdue to amongst others heightened physiological responses

6 ISRN Radiology

lack of control over movements and magnetic materials nec-essary for diagnosis and treatment [4] In general the impactof artifacts may be reduced or controlled to some extentthrough careful preparation and experimental design andexclusion of the most affected data may facilitate analysis andinterpretation However it is impossible to exploit clinicaldata fully without advanced approaches to signal processingthat can reliably separate brain signals and artifacts (Figure 2illustrates the concepts discussed in this section)

41 Physiological Artifacts Arguably the two most dominantartifact sources are the heartbeat and eye blink Althoughthese physiological artifacts can have rather large amplitudessignal separation is challenging as the artificial signals mayresemble activity originating from neuronal sources Regard-ing the eye normal automatic blinking occurs between about5 and 20 times per minute and causes magnetic fields overpredominately frontotemporal regions that can be muchlarger than the magnetic fields of spontaneous and evokedbrain activity The primary artifact is generated by currentflow between the eye and scalp and by excitation of ocularmuscles controlling eye and eyelid movements The artifacttime course recorded by MEG or EEG during a spontaneousblink corresponds to the eyelid motion reaching a closedposition on average 120ms after the blink is initiated andbefore returning to a fully open position approximately240ms after closure In addition to the direct artifact fieldsblink related neuronal activity lasting up to several tenths ofa second is also often observed over posterior parietal andfrontal cortices as well as other areas of the brain associatedwith higher-order processing [57 58] For artifact correctionone generally obtains an independent record of the artifactby measuring the electrooculogram (EOG) with electrodesplaced close to the eyes

Regarding the heart electrical currents moving throughthe organ during a heartbeat give rise to magnetic fields thatare over the chest between two and three orders of magni-tude larger than the brain magnetic fields The time course ofthe cardiac artifact closely corresponds to the propagation ofelectrical activity through the heart muscle Each heartbeatbegins with an impulse from the sinoatrial node the heartrsquosmain pacemaker activating the upper chambers (atria) Thisactivation is seen as the P wave (duration about 80ms) in theelectrocardiogram (ECG) Next the electrical current flowsdown the heart and activates the lower chambers (ventricles)giving rise to a characteristicwave known as theQRS complex(80ndash120ms) Subsequently the electrical current spreadsback over the ventricles in the opposite direction generatinga recovery wave commonly denoted by the letter T (160ms)Typically these currents flowing in the heart are observableto varying degrees in MEG and EEG recordings causingartifacts that may resemble epileptic spikes and slow waves Itis generally assumed that secondary artifacts due to heart beatrelated blood pulsation body movements (ballistocardio-effects) and susceptibility changes are negligible but mayexist in MEG and EEG recordings if the lungs skin or bloodvessels contain magnetic particles [59 page 162] and [60]For artifact correction one generally obtains an independent

record of the artifact by measuring the ECG with electrodesplaced on both arms close to the wrists

Various approaches have been suggested in the past tocorrect for these artifacts often but not exclusively utilizingequivalent current dipoles to model nonneuronal aspectsof the artifacts for example the retinarsquos electrical charge[61] Despite merits however those methods are increas-ingly being replaced by signal projection-based approachesProjection methods are broadly analogous to obtaining a2-dimensional shadow from a 3-dimensional object wherethe direction of illumination determines which features arevisible in the projection Mathematically the data are repre-sented as points (also called vectors) in a space of dimension119899 equal to the number of measurement channels Each 119899-dimensional vector corresponds to one sampled time pointThen the set of new vectors 119881 is calculated to form a basisfor the unwanted signal features that is the artifact to beremoved In other words the set 119881 spans a subspace that isas far as possible occupied by the artifact but not the neuro-physiological signals of interest An artifact-free (clean)measurement is readily obtained through projection

119872119888

= (1 minus 119881 times 119881+

) times 119872 (6)

where 119872 denotes the measurement matrix of dimension119899 times (number of time-points) 1 denotes the 119899-dimensionalidentity matrix and 119881+ is the pseudoinverse of 119881 which hasdimension 119899 times 119896 where 119896 is less than 119873 [62] Dependenton the artifact to be removed 119896 typically assumes a valuebetween 1 and 3

Perhaps the most powerful method to determine theartifact subspace 119881 is a type of blind source separation(BSS) known as independent component analysis (ICA) ICAis entirely data driven and assumes that the observations119872 constitute a linear superposition (mixing) of statisticallyindependent source signals The number of independentsources 119904 is assumed to be the most equal to the number ofmeasurement channels A frequently quoted analogy is theldquococktail party problemrdquo In this scenario party guests holdmultiple simultaneous conversations recorded using micro-phones located at fixed positions around the room Knowingthe number of speakers (s le number of microphones) ICAallows recovering of the individual conversations by unmix-ing the recordings

119878 = 119880 times119872 (7)

where the number of rows of 119878 is equal to the number ofspeakers Each row of 119878 contains the sound from one speakeronly The number of columns of the un-mixing matrix 119880 isequal to the number of microphones and each element of119880 can be interpreted as a weight determining how much anindividual speakerrsquos voice contributed to a given microphone[63 64] Using this analogy the task is then to identify theldquospeakersrdquo who deliver the artifact and guide them out theroom

Over the last 7 years or so several algorithms have beenpublished in order to address this issue in an automatedfashion The main feature that distinguishes the differentapproaches is whether or not an independent EOG andor

ISRN Radiology 7

EOG

CorrectedOriginal

Source power

Original Corrected

Front

Back

RightLeftSi

gnal

(fT

cm) 300

200100

0minus100

minus200

Time (s)4 5 6 7 8

(a)

(b)

(c)

Figure 2 Artifact removal (a) Shown are data from one channel before (black) and after (red) ICA-based correction for eye-blink artifactsidentified by EOG activity (bottom trace) The inset illustrates the 2-dimensional artifact space used for projection (time-courses and spatialtopographies) The first dimension is dominant Note the artifact has similar features as epileptic discharges in that it involves several MEGchannels has sharp peaks and stands out from ongoing background activity (see also [55]) (b) Equivalent current dipole modeling of anevoked somatosensory response in a child before (left) and after (right) SSS-based movement correction After correction the ECD assumesa physiologically plausible location (courtesy of Elekta Neuromag OY Helsinki) (c) Brain activity can be localized accurately despite strongartifacts caused by DBS electrodes (right) Before beamforming-based correction strong nonphysiological interferences outside the brainare observed (left adapted from [56])

ECG recording is needed for artifact identification Table 2shows a brief overview of some of the relevant studies

A recent study in 26 healthy volunteers compared variousartifact detection metrics as well as different flavors of ICAand other blind source separation techniques [70] In gen-eral the authors found automated artifact removal methodspracticable where satisfactory results can be obtained inmostcases irrespective of the detailed nature of the algorithmDependent on type of artifact and data to be cleaned how-ever a combination of approaches employing specific detec-tion metrics might be needed in order to reduce unwantedsignals to an acceptable level It is noteworthy to know thatvalidation has so far been performed in healthy volunteersonly making it difficult to judge when fully automatedmethods will meet clinical standards

42 Head Movement Head or more general subject move-ment during signal acquisition can significantly affect theusability of MEG data due nonneuronal signals arising frommuscle activity artifacts caused by motion related scannervibrations and changes in spatial reference impacting sub-sequent source localization Typically healthy adult patientsare highly motivated and maintain a constant head positionwithin about 2mm standard deviation of the measured headpositions in particular when scans are short Movementrelated artifacts however might increase substantially incertain patient populations and in pediatric measurements[4] Early approaches aimed at preventing head movement asfar as possible with deflatable support cushions bite-bars orother constraints In contrast modern approaches increasingrely on continuous position monitoring in conjunction with

computational methods to correct for head movementsduring data preprocessing

Arguably the most advanced approaches utilize signal-space-separation (SSS) This technique is based on funda-mental properties of quasi-static magnetic fields implyingthat the measurement geometry can be divided into twosource volumes with respect to the sensor array The internalvolume is a sphere whose radius is defined as the distancefrom the head origin to the nearest MEG sensorThe externalvolume is defined as the outside of a spherewhere the radius isthe distance from the head origin to the farthest MEG sensorThen themagnetic fieldmeasured by the sensors (119872) locatedin the current-free space between the internal and externalvolumes can be represented as

119872 = 119872int +119872ext = 119867 times 119860 = [119867int 119867ext] [119860 int119860ext

]

= 119867int times 119860 int + 119867ext times 119860ext

(8)

where 119867 depends on the sensor array and its location withrespect to the two source volumes In contrast the multipolemoments 119860 are device and geometry independent and canbe estimated from the original measurement 119872 Only theinternal terms in the formula above are of interest as they areassumed to represent sources in the brain and a signal-spaceseparated (filtered) signal is obtained by omitting the externalterms thereby suppressing artifacts originating far away fromthe MEG sensors Critically the device independency of 119860allows correction of head movements according to the fol-lowing

119872119904

= 119867119904

times 119860 int (9)

8 ISRN Radiology

Table 2 Overview of some of the relevant ICA-based methods for ocular andor cardiac artifact removal in chronological order

First author Artifact Electrode Sample Comment

Rong [65] Ocular cardiac No 5 One of the first automated removal methods requires a predeterminedartifact template

Okada [66] Ocular Yes 7 Reliable performance for blinks occurring in rapid successionMantini [67] Ocular cardiac No 12 Requires complex parameter tuning can correct environmental artifactsDammers [68] Ocular cardiac Yes 6 Uses signal amplitudes and phases can correct for muscle artifactsKlados [69] Ocular Yes 27 Hybrid regression-ICA approach available as open sourceElectrode indicates whether an electrooculogram andor electrocardiogram measurement is required Sample number of healthy volunteers used for valida-tion

where 119867119904

represents a virtual sensor array locked to thesubjectrsquos head In other words one can calculate the signalsrecorded by a sensor array with respect to where the head isstationary provided that a continuousmovementmonitoringmethod is available [71 72] A temporal extension of thisapproach known as tSSS attempts to identify residual signalscommon to 119872int and 119872ext Such signals typically but notnecessarily represent artifacts originating between the insideand outside volumes and once identified can be removedfrom119872int using a projection as described previously [73]

The SSS approach has been tested extensively wherethe suppression of artifacts originating at a large distance(gt1m) from the scanner matches the theory-based expec-tation [74 75] Note that suppression of nearfields such asocular and cardiac artifacts can be unsatisfactory Excellentresults however have been obtained for tSSS in the case ofspeech artifacts where the movements of tongue lips andjaw and the activity of facial muscles produce magnetic fieldsinterfering with the signals [76] This appears to hold alsofor magnetic fillings and other dental implants significantlyincreasing the number of cases that can be successfullyscanned with MEG

Although movement correction based on SSS has beenvalidated using artificial sources relatively few studies haveinvestigated the effects of head motion in vivo Initiallythe location error of the early somatosensory response toelectrical median nerve stimulation was calculated in a singlesubject making controlled headmovements [77]The authorsreported a 3-fold reduction in localization error comparedto raw data when using tSSS-based movement correction forhead shifts up to 3 cmThe limit of 3 cm is broadly consistentwith a single-case study reporting consistent localization ofepileptic activity when correcting for head movement duringseizure (23 arc length) whereas activity located in differentsublobar regions without correction [78] (Note the MEGresults did not change patientmanagement see Section 51 fora discussion of MEG in epilepsy research)

A recent study systematically investigated the effectsof head motion in a larger sample size [79] The authorsrecorded the neuronal response to auditory and somatosen-sory stimuli in 20 healthy volunteers performing controlledhead movements Moreover magnetized particles wereattached to the subjectrsquos head to simulate nearby interferencesources and a phantom head containing current dipoles wasused for cross validation Confirming and extending previousfindings the results showed that source locations can be

recovered reliably with SSS (error about 5mm) for headshifts not exceeding 3 cm Also using tSSS to remove artifactsdue to nearby sources did not affect the reference dataThe authors pointed out that auditory evoked responsesappeared to be more affected by continuous movement thansomatosensory responses implying that the outcomes ofmotion correction have to be judged in a context dependentmanner

43 Implanted Materials Implanted metal containing ob-jects such as electrodes hydrocephalus shunts or ferromag-netic remnants of neurosurgical operations pose particu-larly high challenges for MEG artifact correction as non-physiological magnetic fields are generated ldquoin siturdquo directlyinterfering with the brain signals The artifacts become evenworse when electrically active devices are involved such asbrain stimulators or pacemakers There is currently no uni-versally applicable method available that corrects for thesetypes of artifacts however promising results are beginningto emerge Two examples may suffice for illustration here

One study used MEG to record epileptic discharges inpatients carrying a vagus nerve stimulator which has provenbeneficial to some epileptic patients considered unsuitablecandidates for resective surgery [80] Using tSSS to clean thedata the authors were able to obtain interpretable signals in 7out of 10 patients The clean signals were subsequently mod-eled using equivalent dipoles in order to localize interictalspikes As a consequence of the MEG results managementchanged to invasive video-EEG (IVEM) in two patientsFurthermore the MEG findings supported the proposedmanagement in four patients demonstrating the feasibility ofMEG in patients with a nerve stimulator

Another study recorded simultaneous MEG and localfield potentials in 17 subjects with Parkinsonrsquos disease (PD)carrying a deep brain stimulator (DBS) to alleviate symptomscaused by degrading motor control [81] Using beamform-ing the authors were able to suppress the high-amplitudeartifacts caused by the DBS wire and electrode and extractinterpretable virtual electrode time-series Subsequent time-frequency analysis pointed to specific synchronization in thegamma band (60ndash90Hz) that stimulates movement throughmodulatory effects in the basal ganglia cortical networkAlthough not of direct clinical relevance this finding is ofimportance for a better understanding of motor control inPD

ISRN Radiology 9

5 Clinical Application Epilepsy

The most well-established clinical application of MEG ispresurgical evaluation in epilepsy Epilepsy is a diverse set ofneurological disorders characterized by the tendency to haverecurring seizures A seizure is the response to an abnormalelectrical discharge in the brain and it describes a varietyof behaviors such as psychic aberrations (eg a sense ofdeja vu) abnormal sensations (eg sensing an intense smell)impairments in speech or speech comprehensions and coor-dinated and involuntary movements Both the nature andseverity of seizure symptoms are dependent on which partsof the brain are affected by the abnormal discharges Seizuresof low severity leave consciousness unimpaired whereas themost severe seizuresmay become life-threatening and requireinstant medical attention [82 83]

Often it is unknown what causes abnormal electricaldischarges however certain risk factors have been identifiedsuch as brain lesions disorders of themetabolic system braininfections and drug and substance abuse amongst others Aseizure typically lasts up to several minutes implying that thetime period that exists between seizures (interictal period)accounts for the majority of a patientrsquos life with epilepsyAlthough most patients may appear symptom free duringinterictal periods abnormal dischargesmay be presentwithinthe brain The anatomic origin of such discharges is oftenalthough not always close to the anatomic origin of seizure-related (ictal) discharges implying a clinical relevance ofinterictal epileptic activity [84]

In industrialized countries epilepsy has an estimatedprevalence of 4ndash10 per 1000 and an incidence of 40ndash70 per100000 and year causing substantial medical expendituresThe disorder is usually controlled but not cured withantiepileptic drugs resulting in seizure freedom in about60 of patients The remaining 40 of affected individualshowever have to be considered pharmacoresistant [85]Surgery is an alternative option in these cases if the seizurescan be traced to abnormal electrical discharges in oneor more clearly delineated brain areas Then the removalor disconnection of the epileptogenic tissues can entail highrates of success Epilepsy surgery however faces substan-tial challenges The ultimate goal is improvement of qual-ity of life while avoiding cognitive or other impairmentscaused by damage to essential functional areas This requiresa substantial multitechnological and possibly invasive pre-operative workup aimed at both the accurate localization ofthe epileptic foci and mapping of nearby essential functionalareas where only the combined results can enable successfulsurgery [86]

51 Localization of Epileptic Foci The potential of MEGto localize epileptogenic regions was communicated to thewider scientific audience over 25 years ago [87] Efforts atconfirming the accuracy of MEG source localizations havebeen addressed from various indirect and direct approachesIndirect validation comes predominantly from studies show-ing colocalization of the magnetic source estimates withknown epileptogenic tissues apparent on structural and func-tional MR imaging or confirmed with invasive intracranial

EEG (IC-EEG) and successful surgical outcomes In contrastthe direct methods reflect studies utilizing simultaneousMEG and IC-EEG recordings or simulated epileptic dis-charges generated by implanted dipoles [85 88] Similar toother non-invasive approaches to preoperative evaluation aconsiderable amount of the MEG work has been based oninterictal discharges (an example is shown in Figure 3(a)[55 89ndash91]) However the importance to determine thelocation of the onset of ictal activity associated with clinicalseizures has prompted researcher to address methodologicalchallenges and validate ictal MEG [92]

In one of the first larger studies of its kind successfulictal MEG recordings were made in 6 out of 20 patientswith intractable complex partial seizures [94] All patientshad neocortical epilepsies and data collection took placeover a span of 3 years In order to capture ictal-onsetactivity a physician remained in the magnetically shieldedroom with each patient and retracted the probe soon after aseizure occurred allowing the patient to move freely Duringrecording a secondphysicianmonitored real-time displays ofthe MEG and EEG waveforms and triggered data collectionwhen abnormal discharges were detected On average theictal scans took several hours and the patients were allowedlight sleep unless it was necessary to reposition the head or ifsleep to wake transitions were felt necessary to obtain ictaldata This allowed the authors to record a few seconds ofictal data although their procedure had a limited capabilityto track seizure spread

The authors used single equivalent current dipole tolocalize sources underlying the MEG field patterns detectedduring ictal and interictal data segments Compared to IC-EEG the authors reported that ictalMEGprovided localizinginformation that was superior to interictal MEG in three ofthe six patients Localization of ictal onset byMEGwas foundto be equivalent to invasive EEG in five of the six patientsIn one patient ictal-intracranial EEG showed nonlocalizingdiffuse seizure related activity whereas the MEG resultspointed to a circumscribed ictal onset zone in right prefrontalcortices A resection was performed based in part on theMEG findings leaving the patient seizure free and withoutpostoperative neurocognitive deficit over 2 years of followupDespite a small number of ictal recordings the authors rightlyconcluded that ictal MEG might be superior to invasiverecordings under certain conditions

The study furthered two previous investigations in 4patients [95] and 3 patients [92] with epilepsy respectivelyBoth studies concluded that the ictal MEG results were inagreementwith invasive andnoninvasive preoperative assess-ments Also the ictal and interictal MEG source localizationswere topographically very similar These findings howeverwere rather preliminary in nature and insufficient informa-tion was available as to assess the reliability of the utility ofMEG in identification of epileptogenic zones

Further support for the role of ictal MEG in presurgicalevaluation is provided by a recent study in children withintractable complex partial epilepsy [96] Over a span of35 years the authors successfully obtained ictal data in8 pediatric patients who subsequently underwent surgicalresection During MEG recordings the patients were either

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

6 ISRN Radiology

lack of control over movements and magnetic materials nec-essary for diagnosis and treatment [4] In general the impactof artifacts may be reduced or controlled to some extentthrough careful preparation and experimental design andexclusion of the most affected data may facilitate analysis andinterpretation However it is impossible to exploit clinicaldata fully without advanced approaches to signal processingthat can reliably separate brain signals and artifacts (Figure 2illustrates the concepts discussed in this section)

41 Physiological Artifacts Arguably the two most dominantartifact sources are the heartbeat and eye blink Althoughthese physiological artifacts can have rather large amplitudessignal separation is challenging as the artificial signals mayresemble activity originating from neuronal sources Regard-ing the eye normal automatic blinking occurs between about5 and 20 times per minute and causes magnetic fields overpredominately frontotemporal regions that can be muchlarger than the magnetic fields of spontaneous and evokedbrain activity The primary artifact is generated by currentflow between the eye and scalp and by excitation of ocularmuscles controlling eye and eyelid movements The artifacttime course recorded by MEG or EEG during a spontaneousblink corresponds to the eyelid motion reaching a closedposition on average 120ms after the blink is initiated andbefore returning to a fully open position approximately240ms after closure In addition to the direct artifact fieldsblink related neuronal activity lasting up to several tenths ofa second is also often observed over posterior parietal andfrontal cortices as well as other areas of the brain associatedwith higher-order processing [57 58] For artifact correctionone generally obtains an independent record of the artifactby measuring the electrooculogram (EOG) with electrodesplaced close to the eyes

Regarding the heart electrical currents moving throughthe organ during a heartbeat give rise to magnetic fields thatare over the chest between two and three orders of magni-tude larger than the brain magnetic fields The time course ofthe cardiac artifact closely corresponds to the propagation ofelectrical activity through the heart muscle Each heartbeatbegins with an impulse from the sinoatrial node the heartrsquosmain pacemaker activating the upper chambers (atria) Thisactivation is seen as the P wave (duration about 80ms) in theelectrocardiogram (ECG) Next the electrical current flowsdown the heart and activates the lower chambers (ventricles)giving rise to a characteristicwave known as theQRS complex(80ndash120ms) Subsequently the electrical current spreadsback over the ventricles in the opposite direction generatinga recovery wave commonly denoted by the letter T (160ms)Typically these currents flowing in the heart are observableto varying degrees in MEG and EEG recordings causingartifacts that may resemble epileptic spikes and slow waves Itis generally assumed that secondary artifacts due to heart beatrelated blood pulsation body movements (ballistocardio-effects) and susceptibility changes are negligible but mayexist in MEG and EEG recordings if the lungs skin or bloodvessels contain magnetic particles [59 page 162] and [60]For artifact correction one generally obtains an independent

record of the artifact by measuring the ECG with electrodesplaced on both arms close to the wrists

Various approaches have been suggested in the past tocorrect for these artifacts often but not exclusively utilizingequivalent current dipoles to model nonneuronal aspectsof the artifacts for example the retinarsquos electrical charge[61] Despite merits however those methods are increas-ingly being replaced by signal projection-based approachesProjection methods are broadly analogous to obtaining a2-dimensional shadow from a 3-dimensional object wherethe direction of illumination determines which features arevisible in the projection Mathematically the data are repre-sented as points (also called vectors) in a space of dimension119899 equal to the number of measurement channels Each 119899-dimensional vector corresponds to one sampled time pointThen the set of new vectors 119881 is calculated to form a basisfor the unwanted signal features that is the artifact to beremoved In other words the set 119881 spans a subspace that isas far as possible occupied by the artifact but not the neuro-physiological signals of interest An artifact-free (clean)measurement is readily obtained through projection

119872119888

= (1 minus 119881 times 119881+

) times 119872 (6)

where 119872 denotes the measurement matrix of dimension119899 times (number of time-points) 1 denotes the 119899-dimensionalidentity matrix and 119881+ is the pseudoinverse of 119881 which hasdimension 119899 times 119896 where 119896 is less than 119873 [62] Dependenton the artifact to be removed 119896 typically assumes a valuebetween 1 and 3

Perhaps the most powerful method to determine theartifact subspace 119881 is a type of blind source separation(BSS) known as independent component analysis (ICA) ICAis entirely data driven and assumes that the observations119872 constitute a linear superposition (mixing) of statisticallyindependent source signals The number of independentsources 119904 is assumed to be the most equal to the number ofmeasurement channels A frequently quoted analogy is theldquococktail party problemrdquo In this scenario party guests holdmultiple simultaneous conversations recorded using micro-phones located at fixed positions around the room Knowingthe number of speakers (s le number of microphones) ICAallows recovering of the individual conversations by unmix-ing the recordings

119878 = 119880 times119872 (7)

where the number of rows of 119878 is equal to the number ofspeakers Each row of 119878 contains the sound from one speakeronly The number of columns of the un-mixing matrix 119880 isequal to the number of microphones and each element of119880 can be interpreted as a weight determining how much anindividual speakerrsquos voice contributed to a given microphone[63 64] Using this analogy the task is then to identify theldquospeakersrdquo who deliver the artifact and guide them out theroom

Over the last 7 years or so several algorithms have beenpublished in order to address this issue in an automatedfashion The main feature that distinguishes the differentapproaches is whether or not an independent EOG andor

ISRN Radiology 7

EOG

CorrectedOriginal

Source power

Original Corrected

Front

Back

RightLeftSi

gnal

(fT

cm) 300

200100

0minus100

minus200

Time (s)4 5 6 7 8

(a)

(b)

(c)

Figure 2 Artifact removal (a) Shown are data from one channel before (black) and after (red) ICA-based correction for eye-blink artifactsidentified by EOG activity (bottom trace) The inset illustrates the 2-dimensional artifact space used for projection (time-courses and spatialtopographies) The first dimension is dominant Note the artifact has similar features as epileptic discharges in that it involves several MEGchannels has sharp peaks and stands out from ongoing background activity (see also [55]) (b) Equivalent current dipole modeling of anevoked somatosensory response in a child before (left) and after (right) SSS-based movement correction After correction the ECD assumesa physiologically plausible location (courtesy of Elekta Neuromag OY Helsinki) (c) Brain activity can be localized accurately despite strongartifacts caused by DBS electrodes (right) Before beamforming-based correction strong nonphysiological interferences outside the brainare observed (left adapted from [56])

ECG recording is needed for artifact identification Table 2shows a brief overview of some of the relevant studies

A recent study in 26 healthy volunteers compared variousartifact detection metrics as well as different flavors of ICAand other blind source separation techniques [70] In gen-eral the authors found automated artifact removal methodspracticable where satisfactory results can be obtained inmostcases irrespective of the detailed nature of the algorithmDependent on type of artifact and data to be cleaned how-ever a combination of approaches employing specific detec-tion metrics might be needed in order to reduce unwantedsignals to an acceptable level It is noteworthy to know thatvalidation has so far been performed in healthy volunteersonly making it difficult to judge when fully automatedmethods will meet clinical standards

42 Head Movement Head or more general subject move-ment during signal acquisition can significantly affect theusability of MEG data due nonneuronal signals arising frommuscle activity artifacts caused by motion related scannervibrations and changes in spatial reference impacting sub-sequent source localization Typically healthy adult patientsare highly motivated and maintain a constant head positionwithin about 2mm standard deviation of the measured headpositions in particular when scans are short Movementrelated artifacts however might increase substantially incertain patient populations and in pediatric measurements[4] Early approaches aimed at preventing head movement asfar as possible with deflatable support cushions bite-bars orother constraints In contrast modern approaches increasingrely on continuous position monitoring in conjunction with

computational methods to correct for head movementsduring data preprocessing

Arguably the most advanced approaches utilize signal-space-separation (SSS) This technique is based on funda-mental properties of quasi-static magnetic fields implyingthat the measurement geometry can be divided into twosource volumes with respect to the sensor array The internalvolume is a sphere whose radius is defined as the distancefrom the head origin to the nearest MEG sensorThe externalvolume is defined as the outside of a spherewhere the radius isthe distance from the head origin to the farthest MEG sensorThen themagnetic fieldmeasured by the sensors (119872) locatedin the current-free space between the internal and externalvolumes can be represented as

119872 = 119872int +119872ext = 119867 times 119860 = [119867int 119867ext] [119860 int119860ext

]

= 119867int times 119860 int + 119867ext times 119860ext

(8)

where 119867 depends on the sensor array and its location withrespect to the two source volumes In contrast the multipolemoments 119860 are device and geometry independent and canbe estimated from the original measurement 119872 Only theinternal terms in the formula above are of interest as they areassumed to represent sources in the brain and a signal-spaceseparated (filtered) signal is obtained by omitting the externalterms thereby suppressing artifacts originating far away fromthe MEG sensors Critically the device independency of 119860allows correction of head movements according to the fol-lowing

119872119904

= 119867119904

times 119860 int (9)

8 ISRN Radiology

Table 2 Overview of some of the relevant ICA-based methods for ocular andor cardiac artifact removal in chronological order

First author Artifact Electrode Sample Comment

Rong [65] Ocular cardiac No 5 One of the first automated removal methods requires a predeterminedartifact template

Okada [66] Ocular Yes 7 Reliable performance for blinks occurring in rapid successionMantini [67] Ocular cardiac No 12 Requires complex parameter tuning can correct environmental artifactsDammers [68] Ocular cardiac Yes 6 Uses signal amplitudes and phases can correct for muscle artifactsKlados [69] Ocular Yes 27 Hybrid regression-ICA approach available as open sourceElectrode indicates whether an electrooculogram andor electrocardiogram measurement is required Sample number of healthy volunteers used for valida-tion

where 119867119904

represents a virtual sensor array locked to thesubjectrsquos head In other words one can calculate the signalsrecorded by a sensor array with respect to where the head isstationary provided that a continuousmovementmonitoringmethod is available [71 72] A temporal extension of thisapproach known as tSSS attempts to identify residual signalscommon to 119872int and 119872ext Such signals typically but notnecessarily represent artifacts originating between the insideand outside volumes and once identified can be removedfrom119872int using a projection as described previously [73]

The SSS approach has been tested extensively wherethe suppression of artifacts originating at a large distance(gt1m) from the scanner matches the theory-based expec-tation [74 75] Note that suppression of nearfields such asocular and cardiac artifacts can be unsatisfactory Excellentresults however have been obtained for tSSS in the case ofspeech artifacts where the movements of tongue lips andjaw and the activity of facial muscles produce magnetic fieldsinterfering with the signals [76] This appears to hold alsofor magnetic fillings and other dental implants significantlyincreasing the number of cases that can be successfullyscanned with MEG

Although movement correction based on SSS has beenvalidated using artificial sources relatively few studies haveinvestigated the effects of head motion in vivo Initiallythe location error of the early somatosensory response toelectrical median nerve stimulation was calculated in a singlesubject making controlled headmovements [77]The authorsreported a 3-fold reduction in localization error comparedto raw data when using tSSS-based movement correction forhead shifts up to 3 cmThe limit of 3 cm is broadly consistentwith a single-case study reporting consistent localization ofepileptic activity when correcting for head movement duringseizure (23 arc length) whereas activity located in differentsublobar regions without correction [78] (Note the MEGresults did not change patientmanagement see Section 51 fora discussion of MEG in epilepsy research)

A recent study systematically investigated the effectsof head motion in a larger sample size [79] The authorsrecorded the neuronal response to auditory and somatosen-sory stimuli in 20 healthy volunteers performing controlledhead movements Moreover magnetized particles wereattached to the subjectrsquos head to simulate nearby interferencesources and a phantom head containing current dipoles wasused for cross validation Confirming and extending previousfindings the results showed that source locations can be

recovered reliably with SSS (error about 5mm) for headshifts not exceeding 3 cm Also using tSSS to remove artifactsdue to nearby sources did not affect the reference dataThe authors pointed out that auditory evoked responsesappeared to be more affected by continuous movement thansomatosensory responses implying that the outcomes ofmotion correction have to be judged in a context dependentmanner

43 Implanted Materials Implanted metal containing ob-jects such as electrodes hydrocephalus shunts or ferromag-netic remnants of neurosurgical operations pose particu-larly high challenges for MEG artifact correction as non-physiological magnetic fields are generated ldquoin siturdquo directlyinterfering with the brain signals The artifacts become evenworse when electrically active devices are involved such asbrain stimulators or pacemakers There is currently no uni-versally applicable method available that corrects for thesetypes of artifacts however promising results are beginningto emerge Two examples may suffice for illustration here

One study used MEG to record epileptic discharges inpatients carrying a vagus nerve stimulator which has provenbeneficial to some epileptic patients considered unsuitablecandidates for resective surgery [80] Using tSSS to clean thedata the authors were able to obtain interpretable signals in 7out of 10 patients The clean signals were subsequently mod-eled using equivalent dipoles in order to localize interictalspikes As a consequence of the MEG results managementchanged to invasive video-EEG (IVEM) in two patientsFurthermore the MEG findings supported the proposedmanagement in four patients demonstrating the feasibility ofMEG in patients with a nerve stimulator

Another study recorded simultaneous MEG and localfield potentials in 17 subjects with Parkinsonrsquos disease (PD)carrying a deep brain stimulator (DBS) to alleviate symptomscaused by degrading motor control [81] Using beamform-ing the authors were able to suppress the high-amplitudeartifacts caused by the DBS wire and electrode and extractinterpretable virtual electrode time-series Subsequent time-frequency analysis pointed to specific synchronization in thegamma band (60ndash90Hz) that stimulates movement throughmodulatory effects in the basal ganglia cortical networkAlthough not of direct clinical relevance this finding is ofimportance for a better understanding of motor control inPD

ISRN Radiology 9

5 Clinical Application Epilepsy

The most well-established clinical application of MEG ispresurgical evaluation in epilepsy Epilepsy is a diverse set ofneurological disorders characterized by the tendency to haverecurring seizures A seizure is the response to an abnormalelectrical discharge in the brain and it describes a varietyof behaviors such as psychic aberrations (eg a sense ofdeja vu) abnormal sensations (eg sensing an intense smell)impairments in speech or speech comprehensions and coor-dinated and involuntary movements Both the nature andseverity of seizure symptoms are dependent on which partsof the brain are affected by the abnormal discharges Seizuresof low severity leave consciousness unimpaired whereas themost severe seizuresmay become life-threatening and requireinstant medical attention [82 83]

Often it is unknown what causes abnormal electricaldischarges however certain risk factors have been identifiedsuch as brain lesions disorders of themetabolic system braininfections and drug and substance abuse amongst others Aseizure typically lasts up to several minutes implying that thetime period that exists between seizures (interictal period)accounts for the majority of a patientrsquos life with epilepsyAlthough most patients may appear symptom free duringinterictal periods abnormal dischargesmay be presentwithinthe brain The anatomic origin of such discharges is oftenalthough not always close to the anatomic origin of seizure-related (ictal) discharges implying a clinical relevance ofinterictal epileptic activity [84]

In industrialized countries epilepsy has an estimatedprevalence of 4ndash10 per 1000 and an incidence of 40ndash70 per100000 and year causing substantial medical expendituresThe disorder is usually controlled but not cured withantiepileptic drugs resulting in seizure freedom in about60 of patients The remaining 40 of affected individualshowever have to be considered pharmacoresistant [85]Surgery is an alternative option in these cases if the seizurescan be traced to abnormal electrical discharges in oneor more clearly delineated brain areas Then the removalor disconnection of the epileptogenic tissues can entail highrates of success Epilepsy surgery however faces substan-tial challenges The ultimate goal is improvement of qual-ity of life while avoiding cognitive or other impairmentscaused by damage to essential functional areas This requiresa substantial multitechnological and possibly invasive pre-operative workup aimed at both the accurate localization ofthe epileptic foci and mapping of nearby essential functionalareas where only the combined results can enable successfulsurgery [86]

51 Localization of Epileptic Foci The potential of MEGto localize epileptogenic regions was communicated to thewider scientific audience over 25 years ago [87] Efforts atconfirming the accuracy of MEG source localizations havebeen addressed from various indirect and direct approachesIndirect validation comes predominantly from studies show-ing colocalization of the magnetic source estimates withknown epileptogenic tissues apparent on structural and func-tional MR imaging or confirmed with invasive intracranial

EEG (IC-EEG) and successful surgical outcomes In contrastthe direct methods reflect studies utilizing simultaneousMEG and IC-EEG recordings or simulated epileptic dis-charges generated by implanted dipoles [85 88] Similar toother non-invasive approaches to preoperative evaluation aconsiderable amount of the MEG work has been based oninterictal discharges (an example is shown in Figure 3(a)[55 89ndash91]) However the importance to determine thelocation of the onset of ictal activity associated with clinicalseizures has prompted researcher to address methodologicalchallenges and validate ictal MEG [92]

In one of the first larger studies of its kind successfulictal MEG recordings were made in 6 out of 20 patientswith intractable complex partial seizures [94] All patientshad neocortical epilepsies and data collection took placeover a span of 3 years In order to capture ictal-onsetactivity a physician remained in the magnetically shieldedroom with each patient and retracted the probe soon after aseizure occurred allowing the patient to move freely Duringrecording a secondphysicianmonitored real-time displays ofthe MEG and EEG waveforms and triggered data collectionwhen abnormal discharges were detected On average theictal scans took several hours and the patients were allowedlight sleep unless it was necessary to reposition the head or ifsleep to wake transitions were felt necessary to obtain ictaldata This allowed the authors to record a few seconds ofictal data although their procedure had a limited capabilityto track seizure spread

The authors used single equivalent current dipole tolocalize sources underlying the MEG field patterns detectedduring ictal and interictal data segments Compared to IC-EEG the authors reported that ictalMEGprovided localizinginformation that was superior to interictal MEG in three ofthe six patients Localization of ictal onset byMEGwas foundto be equivalent to invasive EEG in five of the six patientsIn one patient ictal-intracranial EEG showed nonlocalizingdiffuse seizure related activity whereas the MEG resultspointed to a circumscribed ictal onset zone in right prefrontalcortices A resection was performed based in part on theMEG findings leaving the patient seizure free and withoutpostoperative neurocognitive deficit over 2 years of followupDespite a small number of ictal recordings the authors rightlyconcluded that ictal MEG might be superior to invasiverecordings under certain conditions

The study furthered two previous investigations in 4patients [95] and 3 patients [92] with epilepsy respectivelyBoth studies concluded that the ictal MEG results were inagreementwith invasive andnoninvasive preoperative assess-ments Also the ictal and interictal MEG source localizationswere topographically very similar These findings howeverwere rather preliminary in nature and insufficient informa-tion was available as to assess the reliability of the utility ofMEG in identification of epileptogenic zones

Further support for the role of ictal MEG in presurgicalevaluation is provided by a recent study in children withintractable complex partial epilepsy [96] Over a span of35 years the authors successfully obtained ictal data in8 pediatric patients who subsequently underwent surgicalresection During MEG recordings the patients were either

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

ISRN Radiology 7

EOG

CorrectedOriginal

Source power

Original Corrected

Front

Back

RightLeftSi

gnal

(fT

cm) 300

200100

0minus100

minus200

Time (s)4 5 6 7 8

(a)

(b)

(c)

Figure 2 Artifact removal (a) Shown are data from one channel before (black) and after (red) ICA-based correction for eye-blink artifactsidentified by EOG activity (bottom trace) The inset illustrates the 2-dimensional artifact space used for projection (time-courses and spatialtopographies) The first dimension is dominant Note the artifact has similar features as epileptic discharges in that it involves several MEGchannels has sharp peaks and stands out from ongoing background activity (see also [55]) (b) Equivalent current dipole modeling of anevoked somatosensory response in a child before (left) and after (right) SSS-based movement correction After correction the ECD assumesa physiologically plausible location (courtesy of Elekta Neuromag OY Helsinki) (c) Brain activity can be localized accurately despite strongartifacts caused by DBS electrodes (right) Before beamforming-based correction strong nonphysiological interferences outside the brainare observed (left adapted from [56])

ECG recording is needed for artifact identification Table 2shows a brief overview of some of the relevant studies

A recent study in 26 healthy volunteers compared variousartifact detection metrics as well as different flavors of ICAand other blind source separation techniques [70] In gen-eral the authors found automated artifact removal methodspracticable where satisfactory results can be obtained inmostcases irrespective of the detailed nature of the algorithmDependent on type of artifact and data to be cleaned how-ever a combination of approaches employing specific detec-tion metrics might be needed in order to reduce unwantedsignals to an acceptable level It is noteworthy to know thatvalidation has so far been performed in healthy volunteersonly making it difficult to judge when fully automatedmethods will meet clinical standards

42 Head Movement Head or more general subject move-ment during signal acquisition can significantly affect theusability of MEG data due nonneuronal signals arising frommuscle activity artifacts caused by motion related scannervibrations and changes in spatial reference impacting sub-sequent source localization Typically healthy adult patientsare highly motivated and maintain a constant head positionwithin about 2mm standard deviation of the measured headpositions in particular when scans are short Movementrelated artifacts however might increase substantially incertain patient populations and in pediatric measurements[4] Early approaches aimed at preventing head movement asfar as possible with deflatable support cushions bite-bars orother constraints In contrast modern approaches increasingrely on continuous position monitoring in conjunction with

computational methods to correct for head movementsduring data preprocessing

Arguably the most advanced approaches utilize signal-space-separation (SSS) This technique is based on funda-mental properties of quasi-static magnetic fields implyingthat the measurement geometry can be divided into twosource volumes with respect to the sensor array The internalvolume is a sphere whose radius is defined as the distancefrom the head origin to the nearest MEG sensorThe externalvolume is defined as the outside of a spherewhere the radius isthe distance from the head origin to the farthest MEG sensorThen themagnetic fieldmeasured by the sensors (119872) locatedin the current-free space between the internal and externalvolumes can be represented as

119872 = 119872int +119872ext = 119867 times 119860 = [119867int 119867ext] [119860 int119860ext

]

= 119867int times 119860 int + 119867ext times 119860ext

(8)

where 119867 depends on the sensor array and its location withrespect to the two source volumes In contrast the multipolemoments 119860 are device and geometry independent and canbe estimated from the original measurement 119872 Only theinternal terms in the formula above are of interest as they areassumed to represent sources in the brain and a signal-spaceseparated (filtered) signal is obtained by omitting the externalterms thereby suppressing artifacts originating far away fromthe MEG sensors Critically the device independency of 119860allows correction of head movements according to the fol-lowing

119872119904

= 119867119904

times 119860 int (9)

8 ISRN Radiology

Table 2 Overview of some of the relevant ICA-based methods for ocular andor cardiac artifact removal in chronological order

First author Artifact Electrode Sample Comment

Rong [65] Ocular cardiac No 5 One of the first automated removal methods requires a predeterminedartifact template

Okada [66] Ocular Yes 7 Reliable performance for blinks occurring in rapid successionMantini [67] Ocular cardiac No 12 Requires complex parameter tuning can correct environmental artifactsDammers [68] Ocular cardiac Yes 6 Uses signal amplitudes and phases can correct for muscle artifactsKlados [69] Ocular Yes 27 Hybrid regression-ICA approach available as open sourceElectrode indicates whether an electrooculogram andor electrocardiogram measurement is required Sample number of healthy volunteers used for valida-tion

where 119867119904

represents a virtual sensor array locked to thesubjectrsquos head In other words one can calculate the signalsrecorded by a sensor array with respect to where the head isstationary provided that a continuousmovementmonitoringmethod is available [71 72] A temporal extension of thisapproach known as tSSS attempts to identify residual signalscommon to 119872int and 119872ext Such signals typically but notnecessarily represent artifacts originating between the insideand outside volumes and once identified can be removedfrom119872int using a projection as described previously [73]

The SSS approach has been tested extensively wherethe suppression of artifacts originating at a large distance(gt1m) from the scanner matches the theory-based expec-tation [74 75] Note that suppression of nearfields such asocular and cardiac artifacts can be unsatisfactory Excellentresults however have been obtained for tSSS in the case ofspeech artifacts where the movements of tongue lips andjaw and the activity of facial muscles produce magnetic fieldsinterfering with the signals [76] This appears to hold alsofor magnetic fillings and other dental implants significantlyincreasing the number of cases that can be successfullyscanned with MEG

Although movement correction based on SSS has beenvalidated using artificial sources relatively few studies haveinvestigated the effects of head motion in vivo Initiallythe location error of the early somatosensory response toelectrical median nerve stimulation was calculated in a singlesubject making controlled headmovements [77]The authorsreported a 3-fold reduction in localization error comparedto raw data when using tSSS-based movement correction forhead shifts up to 3 cmThe limit of 3 cm is broadly consistentwith a single-case study reporting consistent localization ofepileptic activity when correcting for head movement duringseizure (23 arc length) whereas activity located in differentsublobar regions without correction [78] (Note the MEGresults did not change patientmanagement see Section 51 fora discussion of MEG in epilepsy research)

A recent study systematically investigated the effectsof head motion in a larger sample size [79] The authorsrecorded the neuronal response to auditory and somatosen-sory stimuli in 20 healthy volunteers performing controlledhead movements Moreover magnetized particles wereattached to the subjectrsquos head to simulate nearby interferencesources and a phantom head containing current dipoles wasused for cross validation Confirming and extending previousfindings the results showed that source locations can be

recovered reliably with SSS (error about 5mm) for headshifts not exceeding 3 cm Also using tSSS to remove artifactsdue to nearby sources did not affect the reference dataThe authors pointed out that auditory evoked responsesappeared to be more affected by continuous movement thansomatosensory responses implying that the outcomes ofmotion correction have to be judged in a context dependentmanner

43 Implanted Materials Implanted metal containing ob-jects such as electrodes hydrocephalus shunts or ferromag-netic remnants of neurosurgical operations pose particu-larly high challenges for MEG artifact correction as non-physiological magnetic fields are generated ldquoin siturdquo directlyinterfering with the brain signals The artifacts become evenworse when electrically active devices are involved such asbrain stimulators or pacemakers There is currently no uni-versally applicable method available that corrects for thesetypes of artifacts however promising results are beginningto emerge Two examples may suffice for illustration here

One study used MEG to record epileptic discharges inpatients carrying a vagus nerve stimulator which has provenbeneficial to some epileptic patients considered unsuitablecandidates for resective surgery [80] Using tSSS to clean thedata the authors were able to obtain interpretable signals in 7out of 10 patients The clean signals were subsequently mod-eled using equivalent dipoles in order to localize interictalspikes As a consequence of the MEG results managementchanged to invasive video-EEG (IVEM) in two patientsFurthermore the MEG findings supported the proposedmanagement in four patients demonstrating the feasibility ofMEG in patients with a nerve stimulator

Another study recorded simultaneous MEG and localfield potentials in 17 subjects with Parkinsonrsquos disease (PD)carrying a deep brain stimulator (DBS) to alleviate symptomscaused by degrading motor control [81] Using beamform-ing the authors were able to suppress the high-amplitudeartifacts caused by the DBS wire and electrode and extractinterpretable virtual electrode time-series Subsequent time-frequency analysis pointed to specific synchronization in thegamma band (60ndash90Hz) that stimulates movement throughmodulatory effects in the basal ganglia cortical networkAlthough not of direct clinical relevance this finding is ofimportance for a better understanding of motor control inPD

ISRN Radiology 9

5 Clinical Application Epilepsy

The most well-established clinical application of MEG ispresurgical evaluation in epilepsy Epilepsy is a diverse set ofneurological disorders characterized by the tendency to haverecurring seizures A seizure is the response to an abnormalelectrical discharge in the brain and it describes a varietyof behaviors such as psychic aberrations (eg a sense ofdeja vu) abnormal sensations (eg sensing an intense smell)impairments in speech or speech comprehensions and coor-dinated and involuntary movements Both the nature andseverity of seizure symptoms are dependent on which partsof the brain are affected by the abnormal discharges Seizuresof low severity leave consciousness unimpaired whereas themost severe seizuresmay become life-threatening and requireinstant medical attention [82 83]

Often it is unknown what causes abnormal electricaldischarges however certain risk factors have been identifiedsuch as brain lesions disorders of themetabolic system braininfections and drug and substance abuse amongst others Aseizure typically lasts up to several minutes implying that thetime period that exists between seizures (interictal period)accounts for the majority of a patientrsquos life with epilepsyAlthough most patients may appear symptom free duringinterictal periods abnormal dischargesmay be presentwithinthe brain The anatomic origin of such discharges is oftenalthough not always close to the anatomic origin of seizure-related (ictal) discharges implying a clinical relevance ofinterictal epileptic activity [84]

In industrialized countries epilepsy has an estimatedprevalence of 4ndash10 per 1000 and an incidence of 40ndash70 per100000 and year causing substantial medical expendituresThe disorder is usually controlled but not cured withantiepileptic drugs resulting in seizure freedom in about60 of patients The remaining 40 of affected individualshowever have to be considered pharmacoresistant [85]Surgery is an alternative option in these cases if the seizurescan be traced to abnormal electrical discharges in oneor more clearly delineated brain areas Then the removalor disconnection of the epileptogenic tissues can entail highrates of success Epilepsy surgery however faces substan-tial challenges The ultimate goal is improvement of qual-ity of life while avoiding cognitive or other impairmentscaused by damage to essential functional areas This requiresa substantial multitechnological and possibly invasive pre-operative workup aimed at both the accurate localization ofthe epileptic foci and mapping of nearby essential functionalareas where only the combined results can enable successfulsurgery [86]

51 Localization of Epileptic Foci The potential of MEGto localize epileptogenic regions was communicated to thewider scientific audience over 25 years ago [87] Efforts atconfirming the accuracy of MEG source localizations havebeen addressed from various indirect and direct approachesIndirect validation comes predominantly from studies show-ing colocalization of the magnetic source estimates withknown epileptogenic tissues apparent on structural and func-tional MR imaging or confirmed with invasive intracranial

EEG (IC-EEG) and successful surgical outcomes In contrastthe direct methods reflect studies utilizing simultaneousMEG and IC-EEG recordings or simulated epileptic dis-charges generated by implanted dipoles [85 88] Similar toother non-invasive approaches to preoperative evaluation aconsiderable amount of the MEG work has been based oninterictal discharges (an example is shown in Figure 3(a)[55 89ndash91]) However the importance to determine thelocation of the onset of ictal activity associated with clinicalseizures has prompted researcher to address methodologicalchallenges and validate ictal MEG [92]

In one of the first larger studies of its kind successfulictal MEG recordings were made in 6 out of 20 patientswith intractable complex partial seizures [94] All patientshad neocortical epilepsies and data collection took placeover a span of 3 years In order to capture ictal-onsetactivity a physician remained in the magnetically shieldedroom with each patient and retracted the probe soon after aseizure occurred allowing the patient to move freely Duringrecording a secondphysicianmonitored real-time displays ofthe MEG and EEG waveforms and triggered data collectionwhen abnormal discharges were detected On average theictal scans took several hours and the patients were allowedlight sleep unless it was necessary to reposition the head or ifsleep to wake transitions were felt necessary to obtain ictaldata This allowed the authors to record a few seconds ofictal data although their procedure had a limited capabilityto track seizure spread

The authors used single equivalent current dipole tolocalize sources underlying the MEG field patterns detectedduring ictal and interictal data segments Compared to IC-EEG the authors reported that ictalMEGprovided localizinginformation that was superior to interictal MEG in three ofthe six patients Localization of ictal onset byMEGwas foundto be equivalent to invasive EEG in five of the six patientsIn one patient ictal-intracranial EEG showed nonlocalizingdiffuse seizure related activity whereas the MEG resultspointed to a circumscribed ictal onset zone in right prefrontalcortices A resection was performed based in part on theMEG findings leaving the patient seizure free and withoutpostoperative neurocognitive deficit over 2 years of followupDespite a small number of ictal recordings the authors rightlyconcluded that ictal MEG might be superior to invasiverecordings under certain conditions

The study furthered two previous investigations in 4patients [95] and 3 patients [92] with epilepsy respectivelyBoth studies concluded that the ictal MEG results were inagreementwith invasive andnoninvasive preoperative assess-ments Also the ictal and interictal MEG source localizationswere topographically very similar These findings howeverwere rather preliminary in nature and insufficient informa-tion was available as to assess the reliability of the utility ofMEG in identification of epileptogenic zones

Further support for the role of ictal MEG in presurgicalevaluation is provided by a recent study in children withintractable complex partial epilepsy [96] Over a span of35 years the authors successfully obtained ictal data in8 pediatric patients who subsequently underwent surgicalresection During MEG recordings the patients were either

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

8 ISRN Radiology

Table 2 Overview of some of the relevant ICA-based methods for ocular andor cardiac artifact removal in chronological order

First author Artifact Electrode Sample Comment

Rong [65] Ocular cardiac No 5 One of the first automated removal methods requires a predeterminedartifact template

Okada [66] Ocular Yes 7 Reliable performance for blinks occurring in rapid successionMantini [67] Ocular cardiac No 12 Requires complex parameter tuning can correct environmental artifactsDammers [68] Ocular cardiac Yes 6 Uses signal amplitudes and phases can correct for muscle artifactsKlados [69] Ocular Yes 27 Hybrid regression-ICA approach available as open sourceElectrode indicates whether an electrooculogram andor electrocardiogram measurement is required Sample number of healthy volunteers used for valida-tion

where 119867119904

represents a virtual sensor array locked to thesubjectrsquos head In other words one can calculate the signalsrecorded by a sensor array with respect to where the head isstationary provided that a continuousmovementmonitoringmethod is available [71 72] A temporal extension of thisapproach known as tSSS attempts to identify residual signalscommon to 119872int and 119872ext Such signals typically but notnecessarily represent artifacts originating between the insideand outside volumes and once identified can be removedfrom119872int using a projection as described previously [73]

The SSS approach has been tested extensively wherethe suppression of artifacts originating at a large distance(gt1m) from the scanner matches the theory-based expec-tation [74 75] Note that suppression of nearfields such asocular and cardiac artifacts can be unsatisfactory Excellentresults however have been obtained for tSSS in the case ofspeech artifacts where the movements of tongue lips andjaw and the activity of facial muscles produce magnetic fieldsinterfering with the signals [76] This appears to hold alsofor magnetic fillings and other dental implants significantlyincreasing the number of cases that can be successfullyscanned with MEG

Although movement correction based on SSS has beenvalidated using artificial sources relatively few studies haveinvestigated the effects of head motion in vivo Initiallythe location error of the early somatosensory response toelectrical median nerve stimulation was calculated in a singlesubject making controlled headmovements [77]The authorsreported a 3-fold reduction in localization error comparedto raw data when using tSSS-based movement correction forhead shifts up to 3 cmThe limit of 3 cm is broadly consistentwith a single-case study reporting consistent localization ofepileptic activity when correcting for head movement duringseizure (23 arc length) whereas activity located in differentsublobar regions without correction [78] (Note the MEGresults did not change patientmanagement see Section 51 fora discussion of MEG in epilepsy research)

A recent study systematically investigated the effectsof head motion in a larger sample size [79] The authorsrecorded the neuronal response to auditory and somatosen-sory stimuli in 20 healthy volunteers performing controlledhead movements Moreover magnetized particles wereattached to the subjectrsquos head to simulate nearby interferencesources and a phantom head containing current dipoles wasused for cross validation Confirming and extending previousfindings the results showed that source locations can be

recovered reliably with SSS (error about 5mm) for headshifts not exceeding 3 cm Also using tSSS to remove artifactsdue to nearby sources did not affect the reference dataThe authors pointed out that auditory evoked responsesappeared to be more affected by continuous movement thansomatosensory responses implying that the outcomes ofmotion correction have to be judged in a context dependentmanner

43 Implanted Materials Implanted metal containing ob-jects such as electrodes hydrocephalus shunts or ferromag-netic remnants of neurosurgical operations pose particu-larly high challenges for MEG artifact correction as non-physiological magnetic fields are generated ldquoin siturdquo directlyinterfering with the brain signals The artifacts become evenworse when electrically active devices are involved such asbrain stimulators or pacemakers There is currently no uni-versally applicable method available that corrects for thesetypes of artifacts however promising results are beginningto emerge Two examples may suffice for illustration here

One study used MEG to record epileptic discharges inpatients carrying a vagus nerve stimulator which has provenbeneficial to some epileptic patients considered unsuitablecandidates for resective surgery [80] Using tSSS to clean thedata the authors were able to obtain interpretable signals in 7out of 10 patients The clean signals were subsequently mod-eled using equivalent dipoles in order to localize interictalspikes As a consequence of the MEG results managementchanged to invasive video-EEG (IVEM) in two patientsFurthermore the MEG findings supported the proposedmanagement in four patients demonstrating the feasibility ofMEG in patients with a nerve stimulator

Another study recorded simultaneous MEG and localfield potentials in 17 subjects with Parkinsonrsquos disease (PD)carrying a deep brain stimulator (DBS) to alleviate symptomscaused by degrading motor control [81] Using beamform-ing the authors were able to suppress the high-amplitudeartifacts caused by the DBS wire and electrode and extractinterpretable virtual electrode time-series Subsequent time-frequency analysis pointed to specific synchronization in thegamma band (60ndash90Hz) that stimulates movement throughmodulatory effects in the basal ganglia cortical networkAlthough not of direct clinical relevance this finding is ofimportance for a better understanding of motor control inPD

ISRN Radiology 9

5 Clinical Application Epilepsy

The most well-established clinical application of MEG ispresurgical evaluation in epilepsy Epilepsy is a diverse set ofneurological disorders characterized by the tendency to haverecurring seizures A seizure is the response to an abnormalelectrical discharge in the brain and it describes a varietyof behaviors such as psychic aberrations (eg a sense ofdeja vu) abnormal sensations (eg sensing an intense smell)impairments in speech or speech comprehensions and coor-dinated and involuntary movements Both the nature andseverity of seizure symptoms are dependent on which partsof the brain are affected by the abnormal discharges Seizuresof low severity leave consciousness unimpaired whereas themost severe seizuresmay become life-threatening and requireinstant medical attention [82 83]

Often it is unknown what causes abnormal electricaldischarges however certain risk factors have been identifiedsuch as brain lesions disorders of themetabolic system braininfections and drug and substance abuse amongst others Aseizure typically lasts up to several minutes implying that thetime period that exists between seizures (interictal period)accounts for the majority of a patientrsquos life with epilepsyAlthough most patients may appear symptom free duringinterictal periods abnormal dischargesmay be presentwithinthe brain The anatomic origin of such discharges is oftenalthough not always close to the anatomic origin of seizure-related (ictal) discharges implying a clinical relevance ofinterictal epileptic activity [84]

In industrialized countries epilepsy has an estimatedprevalence of 4ndash10 per 1000 and an incidence of 40ndash70 per100000 and year causing substantial medical expendituresThe disorder is usually controlled but not cured withantiepileptic drugs resulting in seizure freedom in about60 of patients The remaining 40 of affected individualshowever have to be considered pharmacoresistant [85]Surgery is an alternative option in these cases if the seizurescan be traced to abnormal electrical discharges in oneor more clearly delineated brain areas Then the removalor disconnection of the epileptogenic tissues can entail highrates of success Epilepsy surgery however faces substan-tial challenges The ultimate goal is improvement of qual-ity of life while avoiding cognitive or other impairmentscaused by damage to essential functional areas This requiresa substantial multitechnological and possibly invasive pre-operative workup aimed at both the accurate localization ofthe epileptic foci and mapping of nearby essential functionalareas where only the combined results can enable successfulsurgery [86]

51 Localization of Epileptic Foci The potential of MEGto localize epileptogenic regions was communicated to thewider scientific audience over 25 years ago [87] Efforts atconfirming the accuracy of MEG source localizations havebeen addressed from various indirect and direct approachesIndirect validation comes predominantly from studies show-ing colocalization of the magnetic source estimates withknown epileptogenic tissues apparent on structural and func-tional MR imaging or confirmed with invasive intracranial

EEG (IC-EEG) and successful surgical outcomes In contrastthe direct methods reflect studies utilizing simultaneousMEG and IC-EEG recordings or simulated epileptic dis-charges generated by implanted dipoles [85 88] Similar toother non-invasive approaches to preoperative evaluation aconsiderable amount of the MEG work has been based oninterictal discharges (an example is shown in Figure 3(a)[55 89ndash91]) However the importance to determine thelocation of the onset of ictal activity associated with clinicalseizures has prompted researcher to address methodologicalchallenges and validate ictal MEG [92]

In one of the first larger studies of its kind successfulictal MEG recordings were made in 6 out of 20 patientswith intractable complex partial seizures [94] All patientshad neocortical epilepsies and data collection took placeover a span of 3 years In order to capture ictal-onsetactivity a physician remained in the magnetically shieldedroom with each patient and retracted the probe soon after aseizure occurred allowing the patient to move freely Duringrecording a secondphysicianmonitored real-time displays ofthe MEG and EEG waveforms and triggered data collectionwhen abnormal discharges were detected On average theictal scans took several hours and the patients were allowedlight sleep unless it was necessary to reposition the head or ifsleep to wake transitions were felt necessary to obtain ictaldata This allowed the authors to record a few seconds ofictal data although their procedure had a limited capabilityto track seizure spread

The authors used single equivalent current dipole tolocalize sources underlying the MEG field patterns detectedduring ictal and interictal data segments Compared to IC-EEG the authors reported that ictalMEGprovided localizinginformation that was superior to interictal MEG in three ofthe six patients Localization of ictal onset byMEGwas foundto be equivalent to invasive EEG in five of the six patientsIn one patient ictal-intracranial EEG showed nonlocalizingdiffuse seizure related activity whereas the MEG resultspointed to a circumscribed ictal onset zone in right prefrontalcortices A resection was performed based in part on theMEG findings leaving the patient seizure free and withoutpostoperative neurocognitive deficit over 2 years of followupDespite a small number of ictal recordings the authors rightlyconcluded that ictal MEG might be superior to invasiverecordings under certain conditions

The study furthered two previous investigations in 4patients [95] and 3 patients [92] with epilepsy respectivelyBoth studies concluded that the ictal MEG results were inagreementwith invasive andnoninvasive preoperative assess-ments Also the ictal and interictal MEG source localizationswere topographically very similar These findings howeverwere rather preliminary in nature and insufficient informa-tion was available as to assess the reliability of the utility ofMEG in identification of epileptogenic zones

Further support for the role of ictal MEG in presurgicalevaluation is provided by a recent study in children withintractable complex partial epilepsy [96] Over a span of35 years the authors successfully obtained ictal data in8 pediatric patients who subsequently underwent surgicalresection During MEG recordings the patients were either

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

ISRN Radiology 9

5 Clinical Application Epilepsy

The most well-established clinical application of MEG ispresurgical evaluation in epilepsy Epilepsy is a diverse set ofneurological disorders characterized by the tendency to haverecurring seizures A seizure is the response to an abnormalelectrical discharge in the brain and it describes a varietyof behaviors such as psychic aberrations (eg a sense ofdeja vu) abnormal sensations (eg sensing an intense smell)impairments in speech or speech comprehensions and coor-dinated and involuntary movements Both the nature andseverity of seizure symptoms are dependent on which partsof the brain are affected by the abnormal discharges Seizuresof low severity leave consciousness unimpaired whereas themost severe seizuresmay become life-threatening and requireinstant medical attention [82 83]

Often it is unknown what causes abnormal electricaldischarges however certain risk factors have been identifiedsuch as brain lesions disorders of themetabolic system braininfections and drug and substance abuse amongst others Aseizure typically lasts up to several minutes implying that thetime period that exists between seizures (interictal period)accounts for the majority of a patientrsquos life with epilepsyAlthough most patients may appear symptom free duringinterictal periods abnormal dischargesmay be presentwithinthe brain The anatomic origin of such discharges is oftenalthough not always close to the anatomic origin of seizure-related (ictal) discharges implying a clinical relevance ofinterictal epileptic activity [84]

In industrialized countries epilepsy has an estimatedprevalence of 4ndash10 per 1000 and an incidence of 40ndash70 per100000 and year causing substantial medical expendituresThe disorder is usually controlled but not cured withantiepileptic drugs resulting in seizure freedom in about60 of patients The remaining 40 of affected individualshowever have to be considered pharmacoresistant [85]Surgery is an alternative option in these cases if the seizurescan be traced to abnormal electrical discharges in oneor more clearly delineated brain areas Then the removalor disconnection of the epileptogenic tissues can entail highrates of success Epilepsy surgery however faces substan-tial challenges The ultimate goal is improvement of qual-ity of life while avoiding cognitive or other impairmentscaused by damage to essential functional areas This requiresa substantial multitechnological and possibly invasive pre-operative workup aimed at both the accurate localization ofthe epileptic foci and mapping of nearby essential functionalareas where only the combined results can enable successfulsurgery [86]

51 Localization of Epileptic Foci The potential of MEGto localize epileptogenic regions was communicated to thewider scientific audience over 25 years ago [87] Efforts atconfirming the accuracy of MEG source localizations havebeen addressed from various indirect and direct approachesIndirect validation comes predominantly from studies show-ing colocalization of the magnetic source estimates withknown epileptogenic tissues apparent on structural and func-tional MR imaging or confirmed with invasive intracranial

EEG (IC-EEG) and successful surgical outcomes In contrastthe direct methods reflect studies utilizing simultaneousMEG and IC-EEG recordings or simulated epileptic dis-charges generated by implanted dipoles [85 88] Similar toother non-invasive approaches to preoperative evaluation aconsiderable amount of the MEG work has been based oninterictal discharges (an example is shown in Figure 3(a)[55 89ndash91]) However the importance to determine thelocation of the onset of ictal activity associated with clinicalseizures has prompted researcher to address methodologicalchallenges and validate ictal MEG [92]

In one of the first larger studies of its kind successfulictal MEG recordings were made in 6 out of 20 patientswith intractable complex partial seizures [94] All patientshad neocortical epilepsies and data collection took placeover a span of 3 years In order to capture ictal-onsetactivity a physician remained in the magnetically shieldedroom with each patient and retracted the probe soon after aseizure occurred allowing the patient to move freely Duringrecording a secondphysicianmonitored real-time displays ofthe MEG and EEG waveforms and triggered data collectionwhen abnormal discharges were detected On average theictal scans took several hours and the patients were allowedlight sleep unless it was necessary to reposition the head or ifsleep to wake transitions were felt necessary to obtain ictaldata This allowed the authors to record a few seconds ofictal data although their procedure had a limited capabilityto track seizure spread

The authors used single equivalent current dipole tolocalize sources underlying the MEG field patterns detectedduring ictal and interictal data segments Compared to IC-EEG the authors reported that ictalMEGprovided localizinginformation that was superior to interictal MEG in three ofthe six patients Localization of ictal onset byMEGwas foundto be equivalent to invasive EEG in five of the six patientsIn one patient ictal-intracranial EEG showed nonlocalizingdiffuse seizure related activity whereas the MEG resultspointed to a circumscribed ictal onset zone in right prefrontalcortices A resection was performed based in part on theMEG findings leaving the patient seizure free and withoutpostoperative neurocognitive deficit over 2 years of followupDespite a small number of ictal recordings the authors rightlyconcluded that ictal MEG might be superior to invasiverecordings under certain conditions

The study furthered two previous investigations in 4patients [95] and 3 patients [92] with epilepsy respectivelyBoth studies concluded that the ictal MEG results were inagreementwith invasive andnoninvasive preoperative assess-ments Also the ictal and interictal MEG source localizationswere topographically very similar These findings howeverwere rather preliminary in nature and insufficient informa-tion was available as to assess the reliability of the utility ofMEG in identification of epileptogenic zones

Further support for the role of ictal MEG in presurgicalevaluation is provided by a recent study in children withintractable complex partial epilepsy [96] Over a span of35 years the authors successfully obtained ictal data in8 pediatric patients who subsequently underwent surgicalresection During MEG recordings the patients were either

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

10 ISRN Radiology

Right

(a)

Right Left Front

(b)

Figure 3 MEG in presurgical evaluation (a) Equivalent current dipole localization (center of circle) for an interictal discharge in a patientwith right frontal epilepsy (b) Distributed source modeling of language function in left hemisphereWernickersquos area for a verb generation task(adapted from [93])

sedated or spontaneously attained stage II sleep because ofpartial sleep deprivation the night before the study Theauthors used single ECD modeling to estimate ictal sourcesbased on high-pass filtered MEG data where the individualfrequency cutoff was determined with short-time Fouriertransform The patient head position was continuouslymonitored and only recordings with less than 5mm headmovement were accepted The concordance between theMEG source localizations and IC-EEG was evaluated atthree levels of anatomical resolution as follows hemisphericlateralization lobar localization and sublobar localization(eg mesial versus lateral)

The authors observed that 5 out of 8 patients had concor-dant interictal discharge and ictal onset MEG localizationsin the same lobe however the ictal source was in generalcloser to the seizure-onset zone as determined by intracranialEEG The authors also reported a high correlation betweenMEG source localization and surgical outcome in particu-lar when the ictal and interictal MEG source localizationsshowed agreement with respect to lateralization and lobarlocalization and only one type of seizure semiology waspresent Acknowledging that the capture of seizures duringMEG recording was challenging the authors concluded thatictal MEG onset can provide useful additional informationfor surgical evaluation of patients with intractable epilepsy

This study is interesting in that the authors comple-mented their ECD calculation which is currently the onlyclinically accepted method with other source estimationssuch as minimum norm and beamforming techniques Thefindings suggest that these methods can yield independentinformation about the size of the initial ictal-onset zoneand its subsequent propagation However the data were toopreliminary in order to reach firm conclusions

Most recently a study reviewed the MEG data of wellover 200 epilepsy patients that had undergone various stagesof clinical assessment and treatment over a total periodof 14 years [97] The authors identified 12 cases who hadsurgery and presented sufficient data to allow retrospectivecomparison of interictal and ictal MEG with intracranial

EEG Spatialtemporal signal separation was used to controlartifacts caused by seizure-related headmovements Epilepticdischarges were modeled using an iterative scheme yieldinga cluster of single equivalent dipoles that best explainedthe seizure-related activity The resultant ECD clusters wereclassified according to two models of different anatomicresolutions in order to compare the source location ofepileptic activity recorded byMEG and intracranial EEGTheauthorsrsquo high-resolution hemisphere-lobe-surface model wasbased on hemisphere lobe and surface of the lobe The low-resolution hemisphere-lobe model was based on hemisphereand lobe only

Taking intracranial EEG as the standard for estimatingictal-onset zones the authors reported high sensitivity (96)and high specificity (90) of ictal MEG at low resolutionAt high spatial resolution a lower sensitivity (71) andspecificity (73) of ictal MEG were observed where thespecificity was about equal to interictal MEG (77) In con-trast the sensitivity of interictal MEG was low (40) Noneof their operated patients had mesial temporal lobe epilepsyhowever the authors found that ictal MEG was equallysensitive and specific on dorsolateral and nondorsolateralneocortical surfaces up to a depth of 4 cm Despite robustdata the authors were rather moderate in concluding thatictalMEGmight have some advantages over interictal record-ings regarding the sensitivity with which epileptic foci aredetected compared to invasive approaches

This study further demonstrates the utility of MEG inevaluation of patients with epilepsy However importantissues remain Three patients out of a larger pool of 22 whounderwent surgery reached Engel classes I (free of disablingseizures) to III (worthwhile improvement) but had no ictalMEG signal suggesting thatMEGalonemight not be sensitiveenough on its own in some cases Challengingly ictal onsetis a highly complex dynamic process where different typesof MEG waveforms in different frequency bands appearafter the initial spike This poses important questions forfuture investigations as to what are the clinical significance ofdifferent MEG waveform types and which source modeling

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

ISRN Radiology 11

approaches are most appropriate to extract clinical informa-tion

To this end challenges will continue to exist Epilepsysimply is a highly complex disorder implying intricate chan-ges in the neuronal dynamics where it is possible that neitherIC-EEG nor surgical seizure-free outcome reliably reflectsepilepsy localizations that define brain areas responsible forthe patientrsquos seizures [85] Notwithstanding such fundamen-tal issues MEG has been recognized as a tool in epilepsydiagnostic mainly as part of the first phase of presurgicalevaluation complementing structural MRI combined videosurface-EEG monitoring and in some cases SPECT andPET [98] In particular there is growing evidence that MEGcan at least in some cases identify epileptogenic lesions notvisible in structural scans [99] It is worth noting that MEGis not a replacement for surface EEG as these technologiesare complementary with respect to sensitivity to epilepticdischarges that is eithermethod candetect spikes not seen bythe other It is not fully resolvedwhat causes these differenceshowever the relative insensitivity of MEG to deeper radialsources (as in temporal lobe epilepsies) paired with a bettersignal-to-noise ratio for sources in superficial neocorticalareas is likely to be explanatory factors [100ndash102]

52 Lateralization of Language Deterioration or even lossof language function is considered a dramatic cognitivecomplication of surgical removal of brain tissue performed toalleviate otherwise intractable neuropathological conditionsTherefore accuracy in the lateralization and localization oflanguage function is of great importance to the clinicianwhenconsidering ablation of tissue The intracarotid amobarbitalprocedure (IAP also known as Wada test) has been thegold standard for lateralization of language dominance beforesurgery since its introduction more than 60 years ago Itis based on injection of amobarbital separately into the leftand right carotid arteries to deactivate one cerebral hemi-sphere at a time Concomitantly the language functions ofthe nonanesthetized hemisphere are tested with a varietyof simple tasks (eg the patient is asked to follow simplecommands name objects or count the days of the week) Ifone hemisphere is language dominant significant languagedisturbances are observed with injection on only that sideUsually the IAP is preceded by an angiogram in order tomapthe cerebral vascular structure forecast how the anestheticwill be distributed in the hemispheres and detect possiblecontraindications precluding the IAP [103 104]

Despite its clinical success however the IAP is an invasiveprocedure involving substantial risk of serious complicationsdue to the angiogram testing and subsequent anesthesia ofthe brain Moreover it is known that the test can pose diffi-culties in interpretation of results because of amongst otherscrossflow of the anesthetic to the other hemisphere insuffi-cient anesthesia of all language areas interactions betweenamobarbital and certain antiepileptic drugs and possibledissociation of language functions within an individualwhere specific functions are represented in each hemisphereSeeking alternative techniques however is not an easy taskbecause of the nature of the comparison The IAP relies

on language testing during a transient inactivation of onehemisphere whereas electrophysiology ormetabolism-basedapproaches assess the spatial and temporal patterns of brainactivity during language-related tasks [105]

One of the largest studies of the concordance of lan-guage dominance between MEG and IAP was conducted in85 patients with intractable seizures evaluated for epilepsysurgery [106]The subjects were required to recognize spokenwords and determine whether each word had been in a listof target words presented before testing Such test of verbalmemory elicits specific event-related fields at long latency(200ndash1000m after stimulus onset) thought to reflect neuronalactivity underlying the execution of higher cognitive func-tions such as recognition judgments syntax and semanticprocessing [107] Using single equivalent dipoles models theauthors observed neuronal sources in the posterior portion ofthe superior temporal gyrus in all patients Secondary sourceswere found in inferior frontal middle and mesial temporalregions to varying degree across subjects Interestingly thesesecondary areas were often found bilaterally even for thosepatients who were unilateral language dominant accordingto the IAP The authors calculated a laterality index based onclusters of dipoles for each subject as follows

LI =119873119877

minus 119873119871

119873119877

+ 119873119871

(10)

where 119873119877

and 119873119871

denote the number of dipole clusters inthe right and left perisylvian regions respectively A lateralityindex between minus01 and 01 was considered to indicatebilateral symmetric activation whereas indices smaller thanminus01 or greater than 01 indicated left or right hemispheredominance Using this index the authors found a strongconcordance between the MEG and IAP results (87)

In themajority of the discordant cases the dipole analysissuggested bilateral language whereas the IAP showed lefthemisphere dominance (64) In order to further assess theclinical usefulness of the MEG approach the author per-formed a separate albeit related analysis by assessing thepotential presence of eloquent cortex in the hemisphereof seizure onset The concordance between MEG and IAPresults was high where both methods determined either thepresence or absence of language function in the affectedhemisphere in 93 of patients (98 sensitivity 83 selec-tivity) An analysis of the discordant cases suggested thatMEG provided a more conservative test for the presenceof language function in the hemisphere subject to surgerywhere 6 of patients presented language function in theaffected side according to the MEG but not IAP

The same experimental approach was used in a repli-cation study performed several years later [108] Basedon a smaller sample of 35 patients with epilepsy andorbrain tumor undergoing presurgical evaluation the authorsreported a concordance of 69 between the MEG and APIresults regarding hemisphere language dominance Regard-ing the presence or absence of language function in theaffected hemisphere the concordance between the MEG andIAP results was comparable to the original study (86 80sensitivity 100 selectivity) An analysis of the discordant

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

12 ISRN Radiology

cases however suggested that the IAP provided a moreconservative test for the presence of language function in thehemisphere subject to surgery where all discordant cases hadlanguage function in the affected side according to the IAPbut not the MEG The authors argued that their results arebroadly consistentwith the original studywhere an unusuallyhigh rate of atypical IAP language cases in this sample werebelieved to explain the noted discrepancies

An alternative experimental approach to assess languagedominance with MEG was recently employed where 63patients with intractable epilepsy brain tumors or vascularlesions were asked to silently read visually presented words[109] The authors used beamforming to extract sourceactivation curves at individual voxels (virtual electrodes)from the sensor data and calculated event-related desynchro-nization for the 120573 (13ndash25Hz) and low 120574 (25ndash50Hz) frequencybands Using a laterality index for their ERD measure theauthors reported a concordance between the MEG and IAPof 81 The authors argued that functional imaging withspatially filtered MEG based on oscillatory changes wascompetitive with conventional ECD approaches with respectto estimating language dominance Interestingly the authorsachieved a high concordance based on activity in frontal lan-guage areas suggesting ERD can probe distributed languagesystems beyond posterior areas that are commonly modeledwith ECD-based methods [110ndash112]

These results have been corroborated by a number ofstudies typically based on smaller sample size investi-gating hemispheric language lateralization and localizationof language within hemispheres (an example is shown inFigure 3(b)) Reliability and concordance of results hold for avariety of task as such as word recognition word categoriza-tion and semantic judgments with both auditory and visualstimuli (for a review see [113]) Taken the evidence togetherone can safely argue thatMEG is highly reliable relative to theIAP in the determination of language dominance and is ableto provide robust intrahemispheric localization of languageAs such the MEG is comparable to fMRI the prime non-invasive tool for presurgical language evaluation

Currently fMRI is more widespread than MEG and hassome advantages due to higher standardization of technologyprotocols and analytical methods [103] In contrast MEGscanning is absolutely risk-free entails the least discomfortand might be faster than fMRI for certain protocols More-over there is some recent evidence pointing toward fMRI-based language lateralization being sensitive to cerebrallesions and to problems in interpreting activation patternsin patients with atypical language representation [114] TheMEG might be more robust in these respects although fur-ther studies are needed to understand the effect of lesions andother factors on localization of language function in order tominimize the risk of physicians and scientistsmisinterpretingresults

6 Clinical Research Autism

Autism is a heterogeneous neurodevelopmental disorder thatis associated with a triad of characteristic symptoms firstimpairments in social interaction manifested by failure to

develop appropriate peer relationships lack of share of enjoy-ment lack of social and emotional reciprocity or impair-ment in the use of multiple nonverbal behavior secondimpairments in communication as manifested by delay inthe development of spoken language impairment in theability to initiate or sustain a conversation or lack of spon-taneous appropriate social play third restricted repetitiveand stereotyped patterns of behavior interests and activitiesas manifested by inflexible adherence to specific nonfunc-tional routines or rituals stereotyped and repetitive motormannerisms (eg hand or finger flapping) or persistentpreoccupation with objects Autism is one of the recognizeddisorders in the autism spectrum ASD for short [115 116]

The disorder becomes apparent within the first threeyears of life and continues throughout adolescence and adult-hood although the manifestations of the condition changeover time and there is marked interindividual phenotypicvariability The diagnosis of ASD is based solely on clinicalassessment of behavior [117] There is no known cure andthere is no apparent unifying mechanism that may underlieASD at the genetic molecular cellular or systems level [118119]Theprevalence ofASD is currently estimated at 6 in 1000On average males are at higher risk for ASD than females bya ratio of about 4 1 Comorbidity with genetic disorders suchas tuberous sclerosis and Fragile X can occur and up to a thirdof individuals with autism suffer from epilepsy [120]

61 Epileptiform Activity in ASD An important issue indevelopmental neurosciences is the contribution of epilepsyto autism spectrum and acquired language disorders Thisimportance derives from the observation that children withLandau-Kleffner syndrome (LKS) ASD or developmentallanguage disorders who show common impairments in gen-erating and decoding speech are at higher risk for epilepsythan children with the fluent albeit typically aberrantlanguage of verbal children with autism [121ndash123] Despiteconsiderable research into this issue there appears to beonly twoMEG studies investigating epileptic brain activity inindividuals with ASD

InitiallyMEGwas used to record the electrophysiologicalresponse during stage III sleep in 50 children with regressiveASD with onset between 20 and 36 months of age (15out of 50 with a clinical seizure disorder) and 6 childrenwith LKS [124] All children with or without clinically rele-vant seizures occasionally demonstrated unusual behaviors(eg rapid blinking unprovoked crying and brief staringspells) which if found in a normal child might conceivablybe interpreted as indicative of subclinical epilepsy Usingequivalent current dipoles to model neuronal sources theauthors found epileptic brain activity in all children withLKS (5 had complex seizures) where generators are locatedpredominantly in left intra- and perisylvian regions

MEG identified abnormal discharges in 82 of thechildren with ASD When epileptic activity was present inthe ASD the same sylvian regions seen in LKS were activein 85 of the cases Moreover 75 of the ASD childrenwith epileptic discharges demonstrated additional nonsylvianzones Interestingly simultaneous EEG revealed epilepticactivity in only 68 of the children with ASD Despite the

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

ISRN Radiology 13

multifocal nature of the epileptic discharges neurosurgicalintervention aimed at control led to a reduction of autisticfeatures and improvement in language skills in 12 of 18 cases

It is generally agreed that this study has provided someevidence that there is a subset of individuals with ASDs whodemonstrate epileptic activity that might be associated withautistic symptoms even in the absence of a clinical seizuredisorder Also MEG appeared to have significantly greatersensitivity to this epileptic activity than simultaneous EEGreinforcing the role MEG plays as a clinical tool The conclu-sions drawn from this study however have been criticized ongrounds of referral bias methodological shortcomings andpossibly conflating regressive autism and LKS [121 125]

Several years later MEG was used to study spontaneousneural activity in 36 children with autism spectrum disorder[126] The patients were referred for full evaluation andworkup studies although none of the subjects had a diagnosisof seizure disorders A visual inspection of the MEG data bya trained examiner blinded to the subjectrsquos clinical historyrevealed specific abnormalities in the form of low amplitudemonophasic and biphasic spikes as well as acute waves inabout 86 of all children with ASD Source estimates basedon equivalent current dipole modeling suggested genera-tors predominately in perisylvian regions Notably epilepticspikes were mostly found in the right but not left hemispherein individuals with Aspergerrsquos syndrome (a subtype withinASD) In contrast simultaneous EEG recordings revealedabnormal activity in only 3 of the cases

This study provides further evidence that subclinicalepilepsy can be detected in many children with ASD usingMEG which appeared to have greater sensitivity than EEGin this study Moreover there was some evidence suggestingthat clinical subtypes within ASD might be distinguishableaccording to laterality of abnormal discharges Further con-clusions however could not be drawn as comparing typicallydeveloping children were not investigated

Note that a recent study used single-photon emissioncomputed tomography (SPECT) EEG and MEG to inves-tigate ictal and interictal discharges in 15 individuals witha diagnosis of ASD and seizure disorder [127] Somewhatunusual the authors did not report their MEG results How-ever the EEG and SPECT localizations of epileptic fociappeared concordant Interestingly the authors found nosignificant relationship between the regions of hypoperfusionas measured by SPECT and autism symptom severity Thismight imply that epilepsy is not a causal factor for ASDRather epileptic discharges simply occur in areas of the brainthat are also functionally or pathologically involved in ASD

62 Face Processing It is well documented that individualswithASD show impairments that are face linked for examplein patterns of eye contact and response to gaze During devel-opment individuals with ASD typically show reduced mem-ory for faces and a deficit in recognizing facially expressedemotion [128] Autism however is also associated with aheightened ability to recognise upside down faces and toextract information from some face features compared to typ-ically developing individuals Such observations support thesuggestion that individuals with autism attend to individual

features rather than process faces as a whole [129 130] Mostof the neuroimaging literature on face processing in autismhas concentrated on the fusiform face area (FFA) locatedin ventral occipitotemporal cortices The FFA is part of thebrainrsquos face processing system and is selectively activated byimages of faces in typically developing subjects It is generallyaccepted that FFA activation is atypical in individuals withASD [131 132] However there is debate as to whether thisis intrinsic or is connected with the details of the task suchas the degree of engagement Despite the relevance of faceprocessing to the study of autism there have been only a fewpublished MEG studies in individuals with ASD so far

MEG was used to study the neuronal response in 12 ableadults with ASD and 22 adult controls performing image cat-egorization and 1-back image memory task [7] The authorsobserved that the response to images of faces in right FFAat about 145ms after stimulus onset was significantly weakerless lateralized (Figure 4(a)) and less affected by memoryrecall in patients compared to typically developing subjectsMoreover an early latency (30ndash60ms) response to faceimages over right anterior temporal regions was observedduring memory encoding but not memory recall in controlsubjects whereas activity was observed during recall but notencoding in patients The authors argued that their indi-viduals with ASD might have developed differently locatedand functionally different extrastriate processing pathwaysThese pathways seem functionally capable of at least someaspects of face processing It is unresolved however how suchprocessing routes relate altered socially linked cognition

Subsequently MEG was used to study face processing in10 children with ASD and 10 mental age and gender matchedchildren who were typically developing boys aged 7ndash12years [8] The authors reported that the response differencesbetween the two groups of children were less marked thanbetween the adult groups and between children and adultsThere were subtle differences however with greater recruit-ment of occipito-temporal cortices in processing nonfacestimuli and a less face specific response in children with ASDcompared to controls Based on these findings the authorshypothesized that the neural mechanisms underlying faceprocessing are less specialized in children than in adultseven in middle childhood where there appears a divergencebetween the normal and abnormal developing face and objectprocessing systems (Figure 4(b))

Most recently MEG was used to record the responseto Mooney stimuli in 13 adult individuals with ASD and16 healthy controls matched for age gender and IQ [133]Mooney images consist of degraded pictures of human faceswhere all shades of gray have been removed leaving thehighlights in white and the shadows rendered in black Mosttypically developing subjects perceive a face when presentedin a Mooney image but fail to do so when the image isinverted [134] The authors observed that individuals withASD showed reduced detection rates during the perception ofupright Mooney stimuli while responses to inverted imageswere in the normal range Concerning the electrophysio-logical level the authors reported that perception of facesinvolved increased high gamma (60ndash120Hz) activity in afronto-parietal network in typically developing subjects

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

14 ISRN Radiology

TD ASD

RightLeft RightLeft

(a)

350300250200150100500

I

II

III

0

1

2

3

4

300 35025015050 200100001234

TDASD

Time (ms)

RMS-

signa

l (fT

cm

) Adult

(b)

Figure 4 MEG in autism research (a) The ECD locations for responses to images of human faces at about 145ms after stimulus onset ina typically developing subject and an individual with ASD (circle indicates the volume conductor sphere) These images illustrate locationsin the right posterior cortices of the generators where on average dipole locations are more lateral in TD compared to ASD (b) Grandroot-mean-square signals following face images The curves have been obtained by summation over all participants within a participantgroup (blue boys with ASD red typically developing boys and stimulus onset at 0) and channels Even in middle childhood the neuralmechanisms underlying face processing are less specialized than in adults (inset) with greater early activation of posterior occipital cortices(I II) and less specific activation of ventral occipitotemporal cortex (III) particularly in boys with ASD

whereas in the ASD group stronger activation was observedin posterior visual regions The authors have argued thattheir data provide novel evidence for the role of gamma-bandactivity in visuoperceptual dysfunctions possibly underlyingimpairments in face processing

63 Summary Autism Beginning with the late 1990s MEGhas been used increasingly as an investigative tool to study theneurophysiological basis of autism spectrum disorders mir-roring a general trend of intensified usage of neuroimagingtechnologies in developmental disorders and pediatrics [3]Over and above epilepsy and face processing research hasbeen directed at characterizing and understanding variousaspects of neural processing which are affected amongst indi-viduals on the autism spectrum such as auditory processing[135] semantic processing [136] and theory-of-mind pro-gressed through study of imitation-related processes [137]The present MEG contributions to autism research howeverare still limited in that studies are rather fragmented interms of experimental design and scientific objectives samplesizes are often typically low subject matching criteria varysubstantially between investigations and comparisons areoften limited to ASD and typically developing populationsrather than pathologies exhibiting some overlap with theautistic spectrum Nonetheless MEG research into autismhas provided insight in line with the more general observa-tion that affected individuals may use different strategies toaccomplish familiar tasks [129 138]

7 Conclusion

Since its introduction a little over 40 years ago magnetoen-cephalography has come a long way being now regarded asone of the most modern imaging tools available to radiolo-gists Many institutions laboratories and hospitals all overthe world are increasingly embracing this noninvasive tech-nology However the role of MEG in diagnosis prognosisand patient treatment is still limited and MEG researchapplications will have to mature before they can become

clinically relevant Moreover there are challenges inherentin the technology that will have to be addressed in the nearfuture MEG is relatively expensive it is reliant on naturalgas resources for normal operation and analytical approacheshave not been standardized yet Despite its disadvantagesMEG is an exquisitely patient friendly tool capable of con-tributing indispensable information about brain dynamicsnot easily obtained with other technologies Ultimately anyneuroimaging technology has inherent limitations and itis likely that a combination of several modalities includingMEG will unveil the most information about the neuronalcorrelates of behavior under normal and pathological con-ditions Thus it appears that MEG is beginning to fulfill thepotential as a clinical tool for a wide range of disorders

Disclosure

Theauthor reports no biomedical financial interests or poten-tial conflict of interests

Acknowledgments

The work was supported by the Department of PsychiatryUniversity of Oxford The author wishes to thank BanuAhtam Morten Kringelbach and Hamid Mohseni for adviceon the paper and the author thanks Clare OrsquoDonoghue forhelp with Figure 1 All work was performed at the OxfordCentre for Human Brain Activity

References

[1] D Cohen ldquoMagnetoencephalography detection of the brainrsquoselectrical activity with a superconducting magnetometerrdquo Sci-ence vol 175 no 4022 pp 664ndash666 1972

[2] E Pataraia C Baumgartner G Lindinger and LDeecke ldquoMag-netoencephalography in presurgical epilepsy evaluationrdquo Neu-rosurgical Review vol 25 no 3 pp 141ndash159 2002

[3] R Paetau ldquoMagnetoencephalography in pediatric neuroimag-ingrdquo Developmental Science vol 5 no 3 pp 361ndash370 2002

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

ISRN Radiology 15

[4] J P Makela N Forss J Jaaskelainen E Kirveskari A Kor-venoja and R Paetau ldquoMagnetoencephalography in neuro-surgeryrdquo Neurosurgery vol 59 no 3 pp 493ndash510 2006

[5] C R McDonald ldquoThe use of neuroimaging to study behaviorin patients with epilepsyrdquo Epilepsy and Behavior vol 12 no 4pp 600ndash611 2008

[6] E S Schwartz J C Edgar W C Gaetz and T P L RobertsldquoMagnetoencephalographyrdquo Pediatric Radiology vol 40 no 1pp 50ndash58 2010

[7] A J Bailey S Braeutigam V Jousmaki and S J SwithenbyldquoAbnormal activation of face processing systems at early andintermediate latency in individuals with autism spectrum dis-order a magnetoencephalographic studyrdquo European Journal ofNeuroscience vol 21 no 9 pp 2575ndash2585 2005

[8] A Kylliainen S Braeutigam J K Hietanen S J Swithenby andA J Bailey ldquoFace- and gaze-sensitive neural responses in chil-dren with autism amagnetoencephalographic studyrdquo EuropeanJournal of Neuroscience vol 24 no 9 pp 2679ndash2690 2006

[9] D Dima S Frangou L Burge S Braeutigam and A C JamesldquoAbnormal intrinsic and extrinsic connectivity within the mag-netic mismatch negativity brain network in schizophrenia apreliminary studyrdquo Schizophrenia Research vol 135 no 1ndash3 pp23ndash27 2012

[10] M Nakamura S Watanabe M Inagaki et al ldquoElectrophysio-logical study of face inversion effects in Williams syndromerdquoBrain and Development vol 35 no 4 pp 323ndash330 2013

[11] R Paetau ldquoMagnetoencephalography in landau-kleffner syn-dromerdquo Epilepsia vol 50 supplement 7 pp 51ndash54 2009

[12] W de Haan W M van der Flier T Koene L L Smits P Schel-tens and C J Stam ldquoDisruptedmodular brain dynamics reflectcognitive dysfunction in Alzheimerrsquos diseaserdquo NeuroImage vol59 no 4 pp 3085ndash3093 2012

[13] C Cheng P Wang W Hsu and Y Lin ldquoInadequate inhibitionof redundant auditory inputs in Alzheimerrsquos disease an MEGstudyrdquo Biological Psychology vol 89 no 2 pp 365ndash373 2012

[14] Y Takei S Kumano S Hattori et al ldquoPreattentive dysfunctionin major depression a magnetoencephalography study usingauditory mismatch negativityrdquo Psychophysiology vol 46 no 1pp 52ndash61 2009

[15] C Dockstader W Gaetz D Cheyne and R Tannock ldquoAbnor-mal neural reactivity to unpredictable sensory events in atten-tion-deficithyperactivity disorderrdquo Biological Psychiatry vol66 no 4 pp 376ndash383 2009

[16] P Helenius M Laasonen L Hokkanen R Paetau and MNiemivirta ldquoImpaired engagement of the ventral attentionalpathway in ADHDrdquo Neuropsychologia vol 49 no 7 pp 1889ndash1896 2011

[17] R Salmelin ldquoClinical neurophysiology of language the MEGapproachrdquo Clinical Neurophysiology vol 118 no 2 pp 237ndash2542007

[18] K Laaksonen L Helle L Parkkonen et al ldquoAlterations in spon-taneous brain oscillations during stroke recoveryrdquo PLoS ONEvol 8 no 4 Article ID e61146 2013

[19] J D Lewine J T Davis E D Bigler et al ldquoObjective docu-mentation of traumatic brain injury subsequent to mild headtrauma multimodal brain imaging with MEG SPECT andMRIrdquo Journal of Head Trauma Rehabilitation vol 22 no 3 pp141ndash155 2007

[20] J D Franzen and T W Wilson ldquoAmphetamines modulate pre-frontal gamma oscillations during attention processingrdquo Neu-roreport vol 23 no 12 pp 731ndash735 2012

[21] M Hamalainen R Hari R J Ilmoniemi J Knuutila and O VLounasmaa ldquoMagnetoencephalographymdashtheory instrumenta-tion and applications to noninvasive studies of the workinghuman brainrdquoReviews ofModern Physics vol 65 no 2 pp 413ndash497 1993

[22] R Hari and N Forss ldquoMagnetoencephalography in the studyof human somatosensory cortical processingrdquo PhilosophicalTransactions of the Royal Society B vol 354 no 1387 pp 1145ndash1154 1999

[23] J Vrba ldquoMultichannel squid biomagnetic systemsrdquoApplicationsof Superconductivity vol 365 pp 61ndash138 1999

[24] J Vrba and S E Robinson ldquoSignal processing in magnetoen-cephalographyrdquoMethods vol 25 no 2 pp 249ndash271 2001

[25] J Vrba ldquoMagnetoencephalography the art of finding a needlein a haystackrdquo Physica C vol 368 no 1ndash4 pp 1ndash9 2002

[26] R Plonsey ldquoThe nature of sources of bioelectric and biomag-netic fieldsrdquo Biophysical Journal vol 39 no 3 pp 309ndash312 1982

[27] J Wu and Y C Okada ldquoGenesis of MEG signals II Effects ofmanipulating voltage- and Ca(2+)-activated K(+) channels onthe magnetic fields produced by the CA3 slice of guinea pigrdquoRecent Advances in Human Neurophysiology vol 1162 pp 38ndash44 1998

[28] J Wu and Y C Okada ldquoGenesis of MEG signals I Effectsof ligand-gated channels on the magnetic fields produced bythe CA3 slice of guinea pigrdquo Recent Advances in HumanNeurophysiology vol 1162 pp 32ndash37 1998

[29] P C Hansen M L Kringelbach and R Salmelin MEG AnIntroduction to Methods Oxford University Press New YorkNY USA 2010

[30] J Gross S Baillet G R Barnes et al ldquoGood practice for con-ducting and reporting MEG researchrdquo Neuroimage vol 65 pp349ndash363 2013

[31] K E Misulis and T Fakhoury Spehlmannrsquos Evoked PotentialPrimer Butterworth-Heinemann Boston Mass USA 3rd edi-tion 2001

[32] E Basar ldquoA reviewof alpha activity in integrative brain functionfundamental physiology sensory coding cognition and pathol-ogyrdquo International Journal of Psychophysiology vol 86 no 1 pp1ndash24 2012

[33] C Tallon-Baudry and O Bertrand ldquoOscillatory gamma activityin humans and its role in object representationrdquo Trends inCognitive Sciences vol 3 no 4 pp 151ndash162 1999

[34] O Bertrand and C Tallon-Baudry ldquoOscillatory gamma activityin humans a possible role for object representationrdquo Interna-tional Journal of Psychophysiology vol 38 no 3 pp 211ndash2232000

[35] P J Uhlhaas and W Singer ldquoNeural synchrony in braindisorders relevance for cognitive dysfunctions and pathophys-iologyrdquo Neuron vol 52 no 1 pp 155ndash168 2006

[36] P J Uhlhaas and W Singer ldquoThe development of neuralsynchrony and large-scale cortical networks during adoles-cence relevance for the pathophysiology of schizophrenia andneurodevelopmental hypothesisrdquo Schizophrenia Bulletin vol 37no 3 pp 514ndash523 2011

[37] A K Engel P Fries and W Singer ldquoDynamic predictionsoscillations and synchrony in top-down processingrdquo NatureReviews Neuroscience vol 2 no 10 pp 704ndash716 2001

[38] M Bartos I Vida and P Jonas ldquoSynaptic mechanisms ofsynchronized gamma oscillations in inhibitory interneuron

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

16 ISRN Radiology

networksrdquoNature Reviews Neuroscience vol 8 no 1 pp 45ndash562007

[39] G Pfurtscheller ldquoFunctional brain imaging based on ERDERSrdquo Vision Research vol 41 no 10-11 pp 1257ndash1260 2001

[40] J D Jackson Classical Electrodynamics John Wiley New YorkNY USA 1962

[41] J Sarvas ldquoBasic mathematical and electromagnetic conceptsof the biomagnetic inverse problemrdquo Physics in Medicine andBiology vol 32 no 1 pp 11ndash22 1987

[42] T S Tian ldquoFunctional data analysis in brain imaging studiesrdquoFrontiers in Psychology vol 1 pp 1ndash11 2010

[43] T E Katila ldquoRound table On the current multipole presenta-tion of the primary current distributionsmdashRome September 151982rdquo Il Nuovo Cimento D vol 2 no 2 pp 660ndash664 1983

[44] S Baillet J CMosher and RM Leahy ldquoElectromagnetic brainmappingrdquo IEEE Signal Processing Magazine vol 18 no 6 pp14ndash30 2001

[45] R D Pascual-Marqui C M Michel and D Lehmann ldquoLowresolution electromagnetic tomography a new method forlocalizing electrical activity in the brainrdquo International Journalof Psychophysiology vol 18 no 1 pp 49ndash65 1994

[46] K Uutela M Hamalainen and E Somersalo ldquoVisualizationof magnetoencephalographic data using minimum currentestimatesrdquo NeuroImage vol 10 no 2 pp 173ndash180 1999

[47] L Stenbacka S Vanni K Uutela and R Hari ldquoComparison ofminimum current estimate and dipole modeling in the analysisof simulated activity in the human visual corticesrdquoNeuroImagevol 16 no 4 pp 936ndash943 2002

[48] S Baillet J C Mosher and R M Leahy ldquoElectromagneticbrain imaging using brainstormrdquo in Proceedings of the 2ndIEEE International Symposium on Biomedical Imaging Macro toNano vol 1 2 pp 652ndash655 April 2004

[49] F Lin T Witzel S P Ahlfors S M Stufflebeam J WBelliveau andM S Hamalainen ldquoAssessing and improving thespatial accuracy in MEG source localization by depth-weightedminimum-norm estimatesrdquo NeuroImage vol 31 no 1 pp 160ndash171 2006

[50] A Gramfort M Kowalski and M Hamalainen ldquoMixed-normestimates for the MEEG inverse problem using acceleratedgradient methodsrdquo Physics in Medicine and Biology vol 57 no7 pp 1937ndash1961 2012

[51] B D Van Veen W Van Drongelen M Yuchtman and ASuzuki ldquoLocalization of brain electrical activity via linearly con-strainedminimum variance spatial filteringrdquo IEEE Transactionson Biomedical Engineering vol 44 no 9 pp 867ndash880 1997

[52] J Gross J Kujala M Hamalainen L Timmermann A Schnit-zler and R Salmelin ldquoDynamic imaging of coherent sourcesstudying neural interactions in the human brainrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 98 no 2 pp 694ndash699 2001

[53] A Hashizume K Iida H Shirozu et al ldquoGradient magnetic-field topography for dynamic changes of epileptic dischargesrdquoBrain Research vol 1144 no 1 pp 175ndash179 2007

[54] J Nenonen ldquoMagnetic Source Imagingrdquo 2007[55] R Enatsu N Mikuni K Usui et al ldquoUsefulness of MEG mag-

netometer for spike detection in patients with mesial temporalepileptic focusrdquo NeuroImage vol 41 no 4 pp 1206ndash1219 2008

[56] H R Mohseni P P Smith C E Parsons et al ldquoMEG can mapshort and long-term changes in brain activity following deep

brain stimulation for chronic painrdquo Plos ONE vol 7 no 6Article ID e37993 2012

[57] R Hari R Salmelin S O Tissari M Kajola and V VirsuldquoVisual stability during eyeblinksrdquoNature vol 367 no 6459 pp121ndash122 1994

[58] T Bardouille T W Picton and B Ross ldquoCorrelates of eyeblinking as determined by synthetic aperture magnetometryrdquoClinical Neurophysiology vol 117 no 5 pp 952ndash958 2006

[59] A Despopoulos and S Silbernagel Color Atlas of PhysiologyThieme Medical Publishers New York NY USA 1991

[60] V Jousmaki and R Hari ldquoCardiac artifacts in magnetoen-cephalogramrdquo Journal of Clinical Neurophysiology vol 13 no2 pp 172ndash176 1996

[61] P Berg and B M Scherg ldquoA multiple source approach to thecorrection of eye artifactsrdquo Electroencephalography and ClinicalNeurophysiology vol 90 no 3 pp 229ndash241 1994

[62] C D Meyer Matrix Analysis and Applied Linear Algebra Bookand Solutions Manual SIAM Philadelphia Pa USA 2001

[63] J Cardoso ldquoBlind signal separation statistical principlesrdquo Pro-ceedings of the IEEE vol 86 no 10 pp 2009ndash2025 1998

[64] C J James and C W Hesse ldquoIndependent component analysisfor biomedical signalsrdquo Physiological Measurement vol 26 no1 pp R15ndashR39 2005

[65] F Rong and J L Contreras-Vidal ldquoMagnetoencephalographicartifact identification and automatic removal based on inde-pendent component analysis and categorization approachesrdquoJournal of Neuroscience Methods vol 157 no 2 pp 337ndash3542006

[66] YOkada J Jung andTKobayashi ldquoAn automatic identificationand removal method for eye-blink artifacts in event-relatedmagnetoencephalographic measurementsrdquo Physiological Mea-surement vol 28 no 12 pp 1523ndash1532 2007

[67] D Mantini R Franciotti G L Romani and V PizzellaldquoImproving MEG source localizations an automated methodfor complete artifact removal based on independent componentanalysisrdquo NeuroImage vol 40 no 1 pp 160ndash173 2008

[68] J Dammers M Schiek F Boers et al ldquoIntegration of ampli-tude and phase statistics for complete artifact removal inindependent components of neuromagnetic recordingsrdquo IEEETransactions onBiomedical Engineering vol 55 no 10 pp 2353ndash2362 2008

[69] M A Klados C Papadelis C Braun and P D Bamidis ldquoREG-ICA a hybridmethodology combining Blind Source Separationand regression techniques for the rejection of ocular artifactsrdquoBiomedical Signal Processing and Control vol 6 no 3 pp 291ndash300 2011

[70] J Escudero R Hornero D Abasolo and A Fernandez ldquoQuan-titative evaluation of artifact removal in real magnetoen-cephalogram signals with blind source separationrdquo Annals ofBiomedical Engineering vol 39 no 8 pp 2274ndash2286 2011

[71] S Taulu and M Kajola ldquoPresentation of electromagnetic mul-tichannel data the signal space separation methodrdquo Journal ofApplied Physics vol 97 no 12 Article ID 124905 2005

[72] S Taulu J Simola and M Kajola ldquoApplications of the signalspace separation methodrdquo IEEE Transactions on Signal Process-ing vol 53 no 9 pp 3359ndash3372 2005

[73] S Taulu and J Simola ldquoSpatiotemporal signal space separationmethod for rejecting nearby interference in MEG measure-mentsrdquo Physics in Medicine and Biology vol 51 no 7 pp 1759ndash1768 2006

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

ISRN Radiology 17

[74] S Taulu M Kajola and J Simola ldquoSuppression of interferenceand artifacts by the signal space separation methodrdquo BrainTopography vol 16 no 4 pp 269ndash275 2004

[75] S Taulu J Simola and M Kajola ldquoClinical applications of thesignal space separation methodrdquo Frontiers in Human BrainTopography vol 1270 pp 32ndash37 2004

[76] M Medvedovsky S Taulu R Bikmullina A Ahonen and RPaetau ldquoFine tuning the correlation limit of spatio-temporalsignal space separation for magnetoencephalographyrdquo Journalof Neuroscience Methods vol 177 no 1 pp 203ndash211 2009

[77] M Medvedovsky S Taulu R Bikmullina and R Paetau ldquoArti-fact and head movement compensation in MEGrdquo NeurologyNeurophysiology and Neuroscience p 4 2007

[78] Y Kakisaka Z I Wang J C Mosher et al ldquoClinical evi-dence for the utility of movement compensation algorithm inmagnetoencephalography successful localization during focalseizurerdquo Epilepsy Research vol 101 no 1-2 pp 191ndash196 2012

[79] J Nenonen J Nurminen D Kicic et al ldquoValidation of headmovement correction and spatiotemporal signal space sepa-ration in magnetoencephalographyrdquo Clinical Neurophysiologyvol 123 no 11 pp 2180ndash2191 2012

[80] E Carrette X De Tiege M O De Beeck et al ldquoMagne-toencephalography in epilepsy patients carrying a vagus nervestimulatorrdquo Epilepsy Research vol 93 no 1 pp 44ndash52 2011

[81] V Litvak A Eusebio A Jha et al ldquoMovement-related changesin local and long-range synchronization in Parkinsonrsquos dis-ease revealed by simultaneous magnetoencephalography andintracranial recordingsrdquo Journal of Neuroscience vol 32 no 31pp 10541ndash10553 2012

[82] The Merck Manual of Medical Information Pocket Book NewYork NY USA 2000

[83] S Karceski ldquoHow to identify the signs and symptoms of partialseizuresrdquo Practical Neurology pp 12ndash13 2008

[84] R C Knowlton and J Shih ldquoMagnetoencephalography inepilepsyrdquo Epilepsia vol 45 no 4 pp 61ndash71 2004

[85] H Stefan S Rampp andRCKnowlton ldquoMagnetoencephalog-raphy adds to the surgical evaluation processrdquo Epilepsy andBehavior vol 20 no 2 pp 172ndash177 2011

[86] J Burch A Marson F Beyer et al ldquoDilemmas in the interpre-tation of diagnostic accuracy studies on presurgical workup forepilepsy surgeryrdquo Epilepsia vol 53 no 8 pp 1294ndash1302 2012

[87] D F Rose P D Smith and S Sato ldquoMagnetoencephalographyand epilepsy researchrdquo Science vol 238 no 4825 pp 329ndash3351987

[88] D F Rose S Sato E Ducla-Soares and C V Kufta ldquoMagne-toencephalographic localization of subdural dipoles in a patientwith temporal lobe epilepsyrdquo Epilepsia vol 32 no 5 pp 635ndash641 1991

[89] C Baumgartner E Pataraia G Lindinger and L DeeckeldquoNeuromagnetic recordings in temporal lobe epilepsyrdquo Journalof Clinical Neurophysiology vol 17 no 2 pp 177ndash189 2000

[90] R Bouet J Jung C Delpuech et al ldquoTowards source volumeestimation of interictal spikes in focal epilepsy using magne-toencephalographyrdquo NeuroImage vol 59 no 4 pp 3955ndash39662012

[91] D Madhavan E Heinrichs-Graham and T W WilsonldquoWhole-brain functional connectivity increases with extendedduration of focal epileptiform activityrdquoNeuroscience Letters vol542 pp 26ndash29 2013

[92] H Stefan S Schneider H Feistel et al ldquoIctal and interictalactivity in partial epilepsy recorded with multichannel magne-toelectroencephalography correlation of electroencephalogra-phyelectrocorticography magnetic resonance imaging singlephoton emission computed tomography and positron emissiontomography findingsrdquo Epilepsia vol 33 no 5 pp 874ndash887 1992

[93] A Ray and S M Bowyer ldquoClinical applications of magne-toencephalography in epilepsyrdquo Annals of Indian Academy ofNeurology vol 13 no 1 pp 14ndash22 2010

[94] D S Eliashiv S M Elsas K Squires I Fried and J Engel JrldquoIctal magnetic source imaging as a localizing tool in partialepilepsyrdquo Neurology vol 59 no 10 pp 1600ndash1610 2002

[95] WW Sutherling P H Crandall and J Engel Jr ldquoThemagneticfield of complex partial seizures agrees with intracranial local-izationsrdquo Annals of Neurology vol 21 no 6 pp 548ndash558 1987

[96] H Fujiwara H M Greiner N Hemasilpin et al ldquoIctal MEGonset source localization compared to intracranial EEG andoutcome improved epilepsy presurgical evaluation in pedi-atricsrdquo Epilepsy Research vol 99 no 3 pp 214ndash224 2012

[97] M Medvedovsky S Taulu E Gaily et al ldquoSensitivity andspecificity of seizure-onset zone estimation by ictal magnetoen-cephalographyrdquo Epilepsia vol 53 no 9 pp 1649ndash1657 2012

[98] A Paulini M Fischer S Rampp et al ldquoLobar localizationinformation in epilepsy patients MEGmdasha useful tool in routinepresurgical diagnosisrdquo Epilepsy Research vol 76 no 2-3 pp124ndash130 2007

[99] M Heers S Rampp H Stefan et al ldquoMEG-based identifica-tion of the epileptogenic zone in occult peri-insular epilepsyrdquoSeizure vol 21 no 2 pp 128ndash133 2012

[100] M Iwasaki E Pestana R C Burgess HO Luders H Shamotoand N Nakasato ldquoDetection of epileptiform activity by humaninterpreters blinded comparison between electroencephalog-raphy and magnetoencephalographyrdquo Epilepsia vol 46 no 1pp 59ndash68 2005

[101] S Knake E Halgren H Shiraishi et al ldquoThe value of mul-tichannel MEG and EEG in the presurgical evaluation of 70epilepsy patientsrdquo Epilepsy Research vol 69 no 1 pp 80ndash862006

[102] A Paulini et al ldquoLobar localisation information comparison ofEEG and MEGrdquo Epilepsia vol 47 pp 105ndash106 2006

[103] B Abou-Khalil ldquoAn update on determination of languagedominance in screening for epilepsy surgery theWada test andnewer noninvasive alternativesrdquo Epilepsia vol 48 no 3 pp442ndash455 2007

[104] S Baxendale ldquoThe Wada testrdquo Current Opinion in Neurologyvol 22 no 2 pp 185ndash189 2009

[105] C Carlson ldquoWada you do for language fMRI and languagelateralizationrdquo Epilepsy Currents vol 10 no 4 pp 86ndash88 2010

[106] A C Papanicolaou P G Simos E M Castillo et al ldquoMag-netocephalography a noninvasive alternative to the Wadaprocedurerdquo Journal of Neurosurgery vol 100 no 5 pp 867ndash8762004

[107] R E Frye R Rezaie and A C Papanicolaou ldquoFunctionalneuroimaging of language using magnetoencephalographyrdquoPhysics of Life Reviews vol 6 no 1 pp 1ndash10 2009

[108] R C Doss W Zhang G L Risse and D L Dickens ldquoLateral-izing language with magnetic source imaging validation basedon theWada testrdquo Epilepsia vol 50 no 10 pp 2242ndash2248 2009

[109] M Hirata T Goto G Barnes et al ldquoLanguage dominanceand mapping based on neuromagnetic oscillatory changes

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

18 ISRN Radiology

comparison with invasive procedures clinical articlerdquo Journalof Neurosurgery vol 112 no 3 pp 528ndash538 2010

[110] R L Billingsley-Marshall T Clear W E Mencl et al ldquoA com-parison of functional MRI and magnetoencephalography forreceptive language mappingrdquo Journal of Neuroscience Methodsvol 161 no 2 pp 306ndash313 2007

[111] K Marinkovic R P Dhond A M Dale M Glessner V Carrand E Halgren ldquoSpatiotemporal dynamics of modality-specificand supramodal word processingrdquo Neuron vol 38 no 3 pp487ndash497 2003

[112] C R McDonald T Thesen D J Hagler Jr et al ldquoDistributedsource modeling of language with magnetoencephalographyapplication to patients with intractable epilepsyrdquo Epilepsia vol50 no 10 pp 2256ndash2266 2009

[113] M Pirmoradi R Beland D K Nguyen B A Bacon and MLassonde ldquoLanguage tasks used for the presurgical assessmentof epileptic patients with MEGrdquo Epileptic Disorders vol 12 no2 pp 97ndash108 2010

[114] J Wellmer B Weber H Urbach J Reul G Fernandez andC E Elger ldquoCerebral lesions can impair fMRI-based languagelateralizationrdquo Epilepsia vol 50 no 10 pp 2213ndash2224 2009

[115] U Frith Autism Explaining the Enigma Blackwell MaldenMass USA 2nd edition 2003

[116] F R Volkmar C Lord A Bailey R T Schultz and A KlinldquoAutism and pervasive developmental disordersrdquo Journal ofChild Psychology and Psychiatry and Allied Disciplines vol 45no 1 pp 135ndash170 2004

[117] S Baron-Cohen and M K Belmonte ldquoAutism a window ontothe development of the social and the analytic brainrdquo AnnualReview of Neuroscience vol 28 pp 109ndash126 2005

[118] E B Caronna J M Milunsky and H Tager-Flusberg ldquoAutismspectrum disorders clinical and research frontiersrdquo Archives ofDisease in Childhood vol 93 no 6 pp 518ndash523 2008

[119] Y Shen K A Dies I A Holm et al ldquoClinical genetic testingfor patientswith autism spectrumdisordersrdquoPediatrics vol 125no 4 pp E727ndashE735 2010

[120] C J Newschaffer L A Croen J Daniels et al ldquoThe epidemi-ology of autism spectrum disordersrdquo Annual Review of PublicHealth vol 28 pp 235ndash258 2007

[121] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I p 558 1999

[122] AM Kanner ldquoCommentary the treatment of seizure disordersand EEG abnormalities in children with autistic spectrumdisorders are we getting ahead of ourselvesrdquo Journal of Autismand Developmental Disorders vol 30 no 5 pp 491ndash495 2000

[123] I Rapin ldquoLanguage heterogeneity and regression in the autismspectrum disordersmdashoverlaps with other childhood languageregression syndromesrdquo Clinical Neuroscience Research vol 6no 3-4 pp 209ndash218 2006

[124] B G R Neville ldquoMagnetoencephalographic patterns of epilep-tiform activity in children with regressive autism spectrumdisordersrdquo Pediatrics vol 104 no 3 I pp 405ndash418 1999

[125] R J Kallen J D Lewine J T Davis et al ldquoA long letter andan even longer reply about autism magnetoencephalographyand electroencephalographyrdquoPediatrics vol 107 no 5 pp 1232ndash1234 2001

[126] J A Munoz-Yunta T Ortiz M Palau-Baduell et al ldquoMagne-toencephalographic pattern of epileptiform activity in children

with early-onset autism spectrum disordersrdquo Clinical Neuro-physiology vol 119 no 3 pp 626ndash634 2008

[127] M Sasaki E Nakagawa K Sugai et al ldquoBrain perfusion SPECTand EEG findings in children with autism spectrum disordersandmedically intractable epilepsyrdquo Brain andDevelopment vol32 no 9 pp 776ndash782 2010

[128] G Dawson S J Webb and J McPartland ldquoUnderstanding thenature of face processing impairment in autism insights frombehavioral and electrophysiological studiesrdquo DevelopmentalNeuropsychology vol 27 no 3 pp 403ndash424 2005

[129] F Happe and U Frith ldquoThe weak coherence account detail-focused cognitive style in autism spectrum disordersrdquo Journalof Autism and Developmental Disorders vol 36 no 1 pp 5ndash252006

[130] K Elgar R Campbell and D Skuse ldquoAre you looking atme Accuracy in processing line-of-sight in Turner syndromerdquoProceedings of the Royal Society B vol 269 no 1508 pp 2415ndash2422 2002

[131] N Hadjikhani R H Joseph J Snyder and H Tager-FlusbergldquoAbnormal activation of the social brain during face perceptionin autismrdquo Human Brain Mapping vol 28 no 5 pp 441ndash4492007

[132] B Jemel LMottron andMDawson ldquoImpaired face processingin autism fact or artifactrdquo Journal of Autism andDevelopmentalDisorders vol 36 no 1 pp 91ndash106 2006

[133] L Sun C Grutzner S Bolte et al ldquoImpaired gamma-bandactivity during perceptual organization in adults with autismspectrum disorders evidence for dysfunctional network activ-ity in frontal-posterior corticesrdquo Journal of Neuroscience vol 32no 28 pp 9563ndash9573 2012

[134] C M Mooney and G A Ferguson ldquoA new closure testrdquoCanadian Journal of Psychology vol 5 no 3 pp 129ndash133 1951

[135] T P L Roberts G L Schmidt M Egeth et al ldquoElectrophys-iological signatures magnetoencephalographic studies of theneural correlates of language impairment in autism spectrumdisordersrdquo International Journal of Psychophysiology vol 68 no2 pp 149ndash160 2008

[136] S Braeutigam S J Swithenby and A J Bailey ldquoContextualintegration the unusual way a magnetoencephalographic studyof responses to semantic violation in individuals with autismspectrum disordersrdquo European Journal of Neuroscience vol 27no 4 pp 1026ndash1036 2008

[137] N Nishitani S Avikainen and R Hari ldquoAbnormal imitation-related cortical activation sequences in Aspergerrsquos syndromerdquoAnnals of Neurology vol 55 no 4 pp 558ndash562 2004

[138] F Happe ldquoAutism cognitive deficit or cognitive stylerdquo Trendsin Cognitive Sciences vol 3 no 6 pp 216ndash222 1999

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom