oulu 2016 d 1392 university of oulu p.o. box 8000 fi-90014...
TRANSCRIPT
UNIVERSITY OF OULU P .O. Box 8000 F I -90014 UNIVERSITY OF OULU FINLAND
A C T A U N I V E R S I T A T I S O U L U E N S I S
Professor Esa Hohtola
University Lecturer Santeri Palviainen
Postdoctoral research fellow Sanna Taskila
Professor Olli Vuolteenaho
University Lecturer Veli-Matti Ulvinen
Director Sinikka Eskelinen
Professor Jari Juga
University Lecturer Anu Soikkeli
Professor Olli Vuolteenaho
Publications Editor Kirsti Nurkkala
ISBN 978-952-62-1413-9 (Paperback)ISBN 978-952-62-1414-6 (PDF)ISSN 0355-3221 (Print)ISSN 1796-2234 (Online)
U N I V E R S I TAT I S O U L U E N S I S
MEDICA
ACTAD
D 1392
ACTA
Vesa Korhonen
OULU 2016
D 1392
Vesa Korhonen
INTEGRATING NEAR-INFRARED SPECTROSCOPY TO SYNCHRONOUS MULTIMODAL NEUROIMAGINGAPPLICATIONS AND NOVEL FINDINGS
UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF MEDICINE;MEDICAL RESEARCH CENTER OULU;OULU UNIVERSITY HOSPITAL
A C T A U N I V E R S I T A T I S O U L U E N S I SD M e d i c a 1 3 9 2
VESA KORHONEN
INTEGRATING NEAR-INFRARED SPECTROSCOPY TO SYNCHRONOUS MULTIMODAL NEUROIMAGINGApplications and novel findings
Academic dissertation to be presented with the assent ofthe Doctora l Train ing Committee of Health andBiosciences of the University of Oulu for public defence inAuditorium 7 of Oulu University Hospital (Kajaanintie50), on 2 December 2016, at 12 noon
UNIVERSITY OF OULU, OULU 2016
Copyright © 2016Acta Univ. Oul. D 1392, 2016
Supervised byDocent Vesa KiviniemiDocent Teemu Myllylä
Reviewed byDocent Jaana HiltunenDoctor Ilkka Nissilä
ISBN 978-952-62-1413-9 (Paperback)ISBN 978-952-62-1414-6 (PDF)
ISSN 0355-3221 (Printed)ISSN 1796-2234 (Online)
Cover DesignRaimo Ahonen
JUVENES PRINTTAMPERE 2016
OpponentDocent Raimo Silvennoinen
Korhonen, Vesa, Integrating near-infrared spectroscopy to synchronousmultimodal neuroimaging. Applications and novel findingsUniversity of Oulu Graduate School; University of Oulu, Faculty of Medicine; MedicalResearch Center Oulu; Oulu University HospitalActa Univ. Oul. D 1392, 2016University of Oulu, P.O. Box 8000, FI-90014 University of Oulu, Finland
Abstract
Brain disorders such as epilepsy, dementia and other mental illnesses induce increasing costs onhealth care systems with aging populations. The most effective treatment of these disorders wouldbe either prevention or intervention of the disorder before irreversible damage develops. However,despite the increased interest in different brain diseases, many of them are still detected too late.One reason for this is the lack of appropriate functional imaging modality that can criticallysample the targeted physiological phenomenon. Furthermore, it has been shown that one imagingmodality is not enough to cover brain functionality properly; a multimodal approach is required.
The main goal of this thesis was to validate near-infrared spectroscopy (NIRS) for brainmeasurement and to integrate it into a multimodal neuroimaging setup that can critically samplebasic human physiological phenomena. A novel key element was the combined use of NIRS withultra-fast magnetic resonance encephalography (MREG), electroencephalography (EEG),continuous non-invasive blood pressure and anesthesia monitoring as a synchronous system. Thisunique multimodal neuroimaging set-up with a new functional magnetic resonance imagingsequence, MREG, can sample human brain physiology at 10 Hz sampling rate withoutcardiorespiratory aliasing.
The implemented setup was successfully used in scanning multiple patient and controlpopulations. With the help of critical sampling rate, non-stationarity between the measured signalsreflecting brain pulsations could be detected. Combined NIRS and EEG showed the capability tomonitor therapeutic opening of the blood-brain barrier during treatment of central nervous systemlymphoma for the first time in humans. Furthermore, our multimodal neuroimaging setup enabledthe mapping of the recently described brain avalanches and glymphatic pulsation mechanisms ofthe brain.
In conclusion, the ultra-fast multimodal laboratory with integrated NIRS offers novel and morecomprehensive views on basic brain physiology. The measures from this thesis also have thepotential to offer new, quantitative biomarkers for the detection of different brain disorders priorto irreversible damage.
Keywords: anesthesia monitoring, BBBD, EEG, fMRI, MREG, multimodal imaging,near-infrared spectroscopy, noninvasive blood pressure measurement
Korhonen, Vesa, Lähi-infrapunaspektroskopian yhdistäminen multimodaaliseenneurokuvantamiseen samanaikaisesti. Sovellukset ja uudet havainnotOulun yliopiston tutkijakoulu; Oulun yliopisto, Lääketieteellinen tiedekunta; Medical ResearchCenter Oulu; Oulun yliopistollinen sairaalaActa Univ. Oul. D 1392, 2016Oulun yliopisto, PL 8000, 90014 Oulun yliopisto
Tiivistelmä
Aivosairaudet kuten epilepsia, dementia ja muut mielenterveyden häiriöt aiheuttavat kasvavissamäärin kuluja ikääntyvien ihmisten terveydenhuollossa. Näiden tautien tehokkain hoitokeinoolisi joko ennaltaehkäisy tai varhainen havaitseminen ennen peruuttamattomien kudosvaurioi-den kehittymistä. Lisääntyneestä kiinnostuksesta huolimatta monet aivosairaudet havaitaan edel-leen liian myöhään. Osasyy tähän on sopivan toiminnallisen kuvausmenetelmän puuttuminen,jolla voitaisiin kuvata haluttu fysiologinen ilmiö riittävän nopeasti. Onkin osoitettu, ettei yksit-täinen kuvausmenetelmä riitä aivojen toiminnan riittävän tarkkaan ymmärtämiseen, vaan siihentarvitaan eri menetelmien yhdistämistä.
Tämän väitöskirjatutkimuksen päätarkoituksena oli arvioida lähi-infrapunaspektroskopian(NIRS) soveltuvuutta aivojen toiminnan mittaamisessa sekä integroida se osaksi multimodaalistaneurokuvantamisjärjestelmää. Uutena elementtinä NIRS:iä käytettiin yhdessä ultranopean mag-neettiresonanssienkefalogrammin (MREG), aivosähkökäyrän (EEG), jatkuva-aikaisen kajoamat-toman verenpaineen mittauksen ja anestesiamonitoroinnin kanssa samanaikaisesti, ajallisestisynkronoituna. Yhdessä uuden toiminnallisen magneettikuvaussekvenssin, MREG:n, kanssa täl-lä ainutlaatuisella multimodaalisella neurokuvantamisjärjestelmällä voidaan kuvata ihmisenaivojen perusfysiologiaa 10 Hz näytteistysnopeudella ilman sydämen sykkeen ja hengityksenlaskostumista.
Toteutetulla multimodaalisella mittausjärjestelmällä tehtiin useita onnistuneita kuvauksia eripotilasryhmillä ja terveillä koehenkilöillä. Kriittisen näytteistämisen ansiosta voitiin havaita epä-stationaarisuutta aivojen pulsaatioita heijastelevien signaalien välillä. NIRS:n ja EEG:n samanai-kainen mittaaminen mahdollisti ensimmäistä kertaa ihmisen veriaivoesteen aukeamisen monito-roinnin keskushermostolymfoomapotilaiden hoidossa. Lisäksi multimodaalinen neurokuvanta-misjärjestelmä mahdollisti hiljattain havaittujen aivojen vyöryjen (engl. avalanches) ja glymfaat-tisten pulsaatioiden kartoittamisen.
Yhteenvetona voidaan todeta, että väitöskirjatyön aikana toteutettu multimodaalinen labora-torio yhdessä NIRS:n kanssa mahdollistaa aivojen perusfysiologian edistyksellisen ja tarkan tut-kimisen. Nyt kehitetyt mittarit saattavat myös tarjota uusia, kvantitatiivisia biomarkkereita eriaivosairauksiin ennen vakavien vaurioiden syntymistä.
Asiasanat: anestesiamonitorointi, BBBD, EEG, fMRI, lähi-infrapunaspektroskopia,MREG, multimodaalinen kuvantaminen, noninvasiivinen verenpaineen mittaus
Vilhelmiinalle ja Outille
8
9
Acknowledgements
The research for this thesis was carried out during the years 2012-2016 at the Oulu
Functional Neuroimaging (OFNI) group, which is a member of the Research Unit
of Medical Imaging, Physics and Technology, Faculty of Medicine, University of
Oulu, in collaboration with Oulu University Hospital and Optoelectronics and
Measurement Techniques Unit.
I want to express my greatest appreciation to my thesis supervisors MD PhD,
Adjunct Professor Vesa Kiviniemi and DSc, Adjunct Professor Teemu Myllylä. You
have both had faith in me from the beginning of my career, which I appreciate
enormously. During this doctoral thesis Vesa has been an excellent principal
supervisor who has always had new inspired ideas that I have put into practice, to
the best of my ability, with other members of the OFNI group. Teemu has been my
closest colleague and an excellent second supervisor. You have helped me in many
different ways in developing and testing devices and analysis methods. Together
we have also made very valuable international research visits, during which we
made measurements using our own measurement devices and established good
working relations with our international colleagues as well.
I’m very thankful for the opportunity to work in the Oulu Functional
Neuroimaging group, which belongs to the Research Unit of Medical Imaging,
Physics and Technology. The help from Professor Osmo Tervonen, Adjunct
Professor Juha Nikkinen, PhD Tuomo Starck, Tuija Keinänen, Jussi Kantola and
Elina Kansanoja has been invaluable, especially in the early days of this project
when I moved permanently from the Optoelectronics and Measurement Techniques
laboratory to work in the hospital. In addition to those mentioned above, I also want
to thank Ville Raatikainen, Janne Kananen, Aleksi Rasila, Niko Huotari, Timo
Tuovinen, Lauri Raitamaa and Taneli Hautaniemi for your help during multimodal
measurements and ordinary work. We had a very nice time in Hawaii with Janne,
Timo and Joonas Autio. Ville, Niko, Janne and Aleksi, you have helped me through
the daily routines and made working a joy. I also want to thank the rest of the OFNI
group for fruitful co-operation.
I also wish to express my gratitude to my co-workers in the Optoelectronics
and Measurement Techniques Unit. Professor Risto Myllylä and DSc Hannu
Sorvoja, you have played an important role in my professional career, starting from
my Master’s Thesis until today. I want to thank Adjunct Professor Alexey Popov,
PhD Alexander Bykov, PhD Mikhail Kirillin, PhD Anton Gorshkov, PhD Ekaterina
Sergeeva and Adjunct Professor Matti Kinnunen as you had a significant role in
10
Study I, as well as Lukasz Surazynski and Aleksandra Zienkiewicz, who had an
important role in Study II. I also want to express my warmest thank to all the people
working in the Optoelectronics and Measurement Techniques Laboratory. I had a
great time working there.
I am very grateful to Jurgen Hennig’s group, especially to Dr Pierre LeVan and
Dr Jakob Assländer, who made the ultra-fast MREG sequence possible for us. It
has had a very significant role in this Doctoral Thesis and will continue to do so in
my future studies.
I am grateful for having had the opportunity to co-operate closely with Oulu
University Hospital, particularly with the departments of Diagnostic Radiology,
Oncology, Anesthesiology and Neurology. My special thanks for the possibility to
measure BBBD treatments go to MD, PhD Matti Isokangas, Adjunct Professor Topi
Siniluoto, MD Eila Sonkajärvi and Professor Outi Kuittinen, as well as to all the
other doctors and nurses taking part in the complex BBBD treatments.
I deeply acknowledge the efforts of all 29 co-authors from many different
countries. Every one of you has had an important role in my studies.
During my Doctoral Thesis work I have had an excellent opportunity to make
a couple of research visits and carry out medical measurements with the multimodal
setup. For this, I want to extend my warmest thanks to Professor Martin Walter,
head of Clinical Affective Neuroimaging Laboratory in Magdeburg, PhD Jurgen
Claassen in Radboud University Nijmegen Medical Center, and Professor Florian
Beissner in Hannover Medical School. You all have been friendly hosts during our
research visits.
I wish to extend my sincere thanks to my Doctoral Thesis Follow-up group
Professor Risto Myllylä, Adjunct Professor Eero Ilkko and Adjunct Professor Eija
Pääkkö.
I wish to express my sincere thanks to pre-examiners Adjunct Professor Jaana
Hiltunen and DSc Ilkka Nissilä for reviewing this thesis.
I want to thank FM Anna Vuolteenaho warmly for carefully revising both the
English and Finnish language of this thesis.
I would like to acknowledge the financial support provided by the University
of Oulu, Oulu University Hospital, MRC Oulu, University of Oulu Graduate
School, SalWe Research Program for Mind and Body (Tekes - the Finnish Funding
Agency for Technology and Innovation grant 1104/10), Academy of Finland grants
123772, 275352 and the following foundations: Instrumentarium Science
Foundation, Oulu University Scholarship Foundation, Tauno Tönning Foundation
and European Cooperation in Science and Technology.
11
My warmest thanks are addressed to my mom who has always been there for
me. Her financial support has also been valuable, especially in the beginning of my
career.
Last but not least, my deepest love and gratitude goes to my beloved wife Outi
and our daughter Vilhelmiina. Outi has given me long-time support and has carried
out most of my family duties whereas Vilhelmiina has always brought sunshine to
my life. I love you both.
Oulu, October 2016 Vesa Korhonen
12
13
Abbreviations
3D Three-dimensional
AP Arterial pressure
BBB Blood-brain barrier
BBBD Blood-brain barrier disruption
BCG Ballistocardiographic
BCI Brain computer interface
BET Brain extraction tool
BH Breath holding
BOLD Blood oxygen level dependent
BP Blood pressure
CA Cerebral autoregulation
CBF Cerebral blood flow
CBV Cerebral blood volume
CMRO2 Cerebral metabolic rate of oxygen
CPS Complex partial seizures
CO2 Carbon dioxide
CSF Cerebrospinal fluid
CT Computed tomography
CW Continuous-wave
DAQ Data acquisition
DC Direct current
DC-EEG Direct current electroencephalography
DMN Default mode network
DMNpcc Posterior cingulate cortex default mode network
DMNvmpf Ventromedial default mode network
DOS Diffuse optical spectroscopy
DOT Diffuse optical topography
DPF Differential pathlength factor
ECG Electrocardiography
EEG Electroencephalography
EtCO2 End-tidal carbon dioxide
FB Full band
FD Frequency-domain
FFT Fast Fourier transform
FIR Finite impulse response
14
fMRI Functional Magnetic Resonance Imaging
FMRIB Oxford centre for functional MRI of the brain
fNIRS Functional near-infrared spectroscopy
FSL FMRIB software library
FWHM Full width at half maximum
GIN Generalized inverse imaging
GLM General linear model
Hb Deoxyhemoglobin
HbO Oxyhemoglobin
HbT Total hemoglobin
HPLED High power light emitting diode
HR Heart rate
HV Hyperventilation
HWM Health and Wellness Measurement
i.a. intra-arterial
IAF Individual alpha-frequency
IC Independent component
ICA Independent component analysis
i.e. id est (~that is)
InI Inverse imaging
i.v. intravenous
LED Light emitting diode
LFO Low frequency oscillation
LKC Lipschitz-Killing curvature
MBLL Modified Beer-Lambert law
MC Monte Carlo
MCFLIRT Motion correction tool for fMRI time series
MDL Minimum description length
MELODIC Multivariate exploratory linear optimized decomposition into
independent components
MNI Montreal neurological institute
MPRAGE Magnetization-prepared rapid acquisition with gradient echo
MR Magnetic resonance
MREG Magnetic resonance encephalography
MRI Magnetic resonance imaging
NIBP Non-invasive blood pressure
NIR Near-infrared
15
NIRS Near-infrared spectroscopy
OPEM Optoelectronics and Measurement Techniques
OFNI Oulu Functional Neuroimaging
PCNSL Primary central nervous system lymphoma
PET Positron emission tomography
Ph.D. Doctor of Philosophy
PICA Probabilistic ICA
PPG Photoplethysmography
PTT Pulse transit time
PWV Pulse wave velocity
RCPS Rapidly secondarily generalized CPS
rCBV Regional cerebral blood volume
RS Resting state
RSFC Resting state functional connectivity
RS-fMRI Resting state fMRI
RSN Resting state network
SaO2 Cerebral oxygenation
SD Standard deviation
SNA Sympathetic nervous activity
SNR Signal-to-noise ratio
SpO2 Pulse oximeter oxygen saturation
TD Time-domain
TR Repetition time
VLF Very low frequency
16
17
List of original publications
This thesis is based on the following publications, which are referred to throughout
the text by their Roman numerals:
I Korhonen V, Myllylä T, Kirillin M, Popov A, Bykov A, Gorshkov A, Sergeeva A, Kinnunen M & Kiviniemi V (2014) Light Propagation in NIR Spectroscopy of the Human Brain. IEEE Journal of Selected Topics in Quantum Electronics 20(2): 7100310.
II Myllylä T, Korhonen V, Surazynski L, Zienkiewicz A, Sorvoja H & Myllylä R (2014) Measurement of cerebral blood flow and metabolism using light-emitting diodes. Measurement 58 (December 2014): 387-393.
III Korhonen V, Hiltunen T, Myllylä T, Wang X, Kantola J, Nikkinen J, Zang Y-F, Levan P & Kiviniemi V (2014) Synchronous multi-scale neuroimaging environment for critically sampled physiological analysis of brain function – Hepta-scan concept. Brain Connectivity 4(9): 677–689.
IV Kiviniemi V, Korhonen V, Kortelainen J, Rytky S, Keinänen (nee Hiltunen) T, Tuovinen T, Isokangas M, Sonkajärvi E, Siniluoto T, Nikkinen J, Alahuhta S, Tervonen O, Turpeenniemi-Hujanen T, Myllylä T, Kuittinen O & Voipio J Real-time monitoring of human blood brain barrier disruption. Manuscript.
These papers are referred to in the text by Roman numerals (I – IV). The current
author is either the first or the second author in all of these publications and had an
active role at all stages of research of this thesis, although the studies are a result of
group effort.
Study I studies light propagation in the human brain by using different
measurement methods and simulations. The results obtained from these
measurements were compared and used to estimate the minimum distance between
the transmitter and receiver in the near-infrared spectroscopy (NIRS) device. In this
study, I had the main responsibility for the NIRS measurements, the NIRS data
analysis and for writing the publication, with active participation of the co-authors.
The Monte Carlo calculation of light attenuation shown in the paper was performed
by Mikhail Kirillin and Ekaterina Sergeeva. The fabrication process of the optical
phantoms, used in this study, was originally developed by Alexey Popov and
Alexander Bykov.
Study II introduces measurement of cerebral blood flow and metabolism using
light-emitting diodes (LEDs) instead of the more commonly used lasers in the
NIRS device. It also goes through major features of the high power LEDs (HPLEDs)
and studies different wavelength combinations used in the NIRS device. I carried
out the measurements and most of the NIRS data analysis (excluding method 2) in
18
this study. I also actively participated in the writing of the article. Method 2, NIRS
data analysis, was performed by Aleksandra Zienkiewicz and Lukasz Surazynski.
Study III introduces a synchronous multi-scale neuroimaging environment for
critically sampled physiological analysis of brain function – the Hepta-scan concept.
The Hepta-scan concept includes functional magnetic resonance imaging (fMRI)
with ultrafast magnetoencephalography (MREG) sequence, NIRS,
electroencephalography (EEG), non-invasive blood pressure (NIBP) measurement
and anesthesia monitor measurements [pulse oximeter oxygen saturation (SpO2),
end-tidal carbon dioxide (EtCO2), electrocardiography (ECG)]. In this study, I
integrated and synchronized NIRS with the multimodal neuroimaging setup. I had
the main responsibility for the multimodal measurements, data collection and data
analysis (excluding EEG data analysis). I also actively participated in the writing
of the manuscript. In this paper Tuija Keinänen performed the EEG data analysis.
In Study IV, human blood-brain barrier disruption (BBBD) was studied in
primary central nervous system lymphoma (PCNSL) patients during chemotherapy
treatment using NIRS and direct current EEG (DC-EEG). In this study, I performed
NIRS measurements and NIRS data analysis. I also actively participated in the
writing of the publication. EEG data analyses were performed by Jukka Kortelainen
and Tuija Keinänen.
19
Table of contents
Abstract
Tiivistelmä
Acknowledgements 9 Abbreviations 13 List of original publications 17 Table of contents 19 1 Introduction 21 2 Review of the literature 23
2.1 Near-infrared spectroscopy in brain studies ............................................ 23 2.1.1 Principles ...................................................................................... 23 2.1.2 Light propagation in human brain ................................................ 24 2.1.3 Different wavelength combinations .............................................. 26 2.1.4 Analysis of NIRS data and software tools .................................... 27 2.1.5 Medical research applications ...................................................... 29 2.1.6 Potential clinical applications ....................................................... 31
2.2 Physiological brain measurements .......................................................... 34 2.2.1 Functional magnetic resonance imaging ...................................... 34 2.2.2 Blood pressure measurements ...................................................... 37 2.2.3 Electroencephalography ............................................................... 38 2.2.4 Physiological signals offered by anesthesia monitoring ............... 39
2.3 NIRS in multimodal brain imaging combinations .................................. 40 2.3.1 NIRS vs. fMRI ............................................................................. 40 2.3.2 NIRS vs. EEG ............................................................................... 42 2.3.3 NIRS vs. NIBP ............................................................................. 44 2.3.4 NIRS/BOLD vs. Cardiorespiratory signals .................................. 45
3 Aims of the study 47 4 Participants and methods 49
4.1 Participants .............................................................................................. 49 4.2 Imaging, measurements and procedure ................................................... 49 4.3 Data pre-processing & analysis ............................................................... 58
5 Results 61 5.1 NIRS measurements ................................................................................ 61
5.1.1 Light propagation (Study I) .......................................................... 61 5.1.2 Different wavelength combinations/flexibility (Study II) ............ 67
5.2 Multimodal measurements (Study III) .................................................... 68
20
5.2.1 Technical realization and integration ............................................ 68 5.2.2 NIRS, EEG and NIBP correlations to MREG DMNvmpf .............. 70 5.2.3 EEG/NIRS/NIBP signal correlations with each other .................. 72
5.3 Real-time monitoring of human BBBD (Study IV) ................................ 74 5.3.1 NIRS ............................................................................................. 74 5.3.2 EEG .............................................................................................. 74
6 Discussion 77 6.1 Properties of the NIRS device ................................................................. 77
6.1.1 Different analysis methods in Study II ......................................... 78 6.2 Feasibility of multimodal imaging .......................................................... 78 6.3 Correlation analysis and non-stationarity of multimodal signals ............ 80 6.4 Real-time monitoring of human BBBD .................................................. 81 6.5 Future directions and technical improvements ........................................ 83
7 Conclusion 85 References 87 Original publications 109
21
1 Introduction
Brain disorders like epilepsy, dementia, affective disorders and other mental
illnesses induce increasing costs on health care systems with aging populations.
The most effective treatment of these disorders would be either prevention or
intervention at the earliest possible phase. However, effective early interventions
would require quantitative biomarkers that enable both the detection of the disorder
prior to irreversible damages and follow-up of treatment effects.
Previously, the imaging of patients focused mainly on detecting anatomical
alterations which often occur late in the disease and involve changes that are
permanent or difficult to remove. However, symptoms of a disease can be detected
at functional level prior to anatomical alterations at a phase that still offers a
window for treatment.
Functional magnetic resonance imaging (fMRI) focusing on spontaneous brain
activity fluctuations can monitor functional connectivity of the brain (Biswal et al. 1995) and has become the fastest growing neuroimaging realm. Recently,
increasing interest has been paid to detecting disease-related changes in resting
state functional connectivity (RSFC) because it offers repeatable and easy imaging
of disease-related alterations even at very early stages of a disease (Filippini et al. 2009, Fox & Greicius 2010, Zhou et al. 2014). However, the fMRI has limitations
such as restrictive usage only inside the MR scanner, sensitivity to artifacts, slow
data sampling and high cost, which considerably restrict its wider usage.
Near-infrared spectroscopy (NIRS) offers an inexpensive, portable, faster and
more versatile functional imaging method. It is increasingly being utilized as a non-
invasive diagnostic tool for studying human physiology. A recent review on the
clinical exploitation of NIRS suggests that it could be usable in stroke risk and
rehabilitation assessment and monitoring (Budohoski et al. 2012), detection of
blood emboli (Oldag et al. 2012), and in migraine and other headache syndromes
related to blood flow vasomotor dysfunction (Liboni et al. 2007, Obrig 2014,
Vernieri et al. 2008). Alzheimer’s dementia and other cognitive diseases could
benefit from correlation of cognitive measures and brain hemodynamics (Arai et al. 2006) whereas epileptic foci detection and seizure monitoring may even show
alterations that precede clinical seizure onset (Nguyen et al. 2013, Slone et al. 2012).
Despite the increased enthusiasm over the capability to detect disease-related
changes in brain activity and blood flow fluctuations, robust individual biomarkers
of disease are still largely missing. The understanding of brain activity fluctuations
22
is still relatively superficial; recent studies on classical blood oxygen level-
dependent (BOLD) signal show non-stationarity in both space and time in the
activity of brain networks (Chang & Glover 2010, Hutchison et al. 2013).
Methodologically, the classic BOLD signal with 2-second repetition time (TR) is
too slowly sampled to critically capture the brain signal fluctuations and suffers
from aliasing of cardiorespiratory signals. These shortcomings impede accurate
analysis of non-stationarity, and faster, critically sampled neuroimaging methods
would thus bring forth unequivocal results.
Because of these desiderata, a multimodal neuroimaging set-up called Hepta-
scan was developed. Hepta-scan includes simultaneously measured ultra-fast
magnetic resonance encephalography (MREG), NIRS, electroencephalography
(EEG), non-invasive blood pressure (NIBP) and anesthesia monitor signals. MREG
is an ultra-fast fMRI imaging sequence that captures BOLD signal 20 times faster
than the conventional fMRI. The Hepta-scan setup can critically sample the major
sources of brain physiological pulsations in millisecond synchrony.
The multimodal approach opened new views into the mechanisms behind slow
brain oscillations. Multimodal data may offer new biomarkers to different diseases,
even at an individual level. One example of a wearable biomarker is the real-time
monitoring of human blood-brain barrier disruption (BBBD) in anesthetized
patients receiving chemotherapy for primary central nervous system lymphoma
(PCNSL).
23
2 Review of the literature
2.1 Near-infrared spectroscopy in brain studies
Jöbsis first introduced diffuse optical spectroscopy (DOS) method for quantifying
concentration changes of blood chromophores in the brain of an animal almost 40
years ago (Jobsis 1977). Since then, DOS has been used and is often referred to as
NIRS, diffuse optical topography (DOT) and today, even as functional NIRS
(fNIRS). Although the name can vary, the method itself is almost the same –
illuminating light onto the scalp, detecting it as it exits the head and using the
absorption spectra of the light-absorbing molecules present in tissue to interpret the
detected light levels as changes in chromophore concentrations (Strangman et al. 2002a). This chapter presents NIRS more in detail.
2.1.1 Principles
When light interacts with biological tissue its transmission depends on a
combination of reflectance, scattering and attenuation properties (Jobsis 1977). The
near-infrared region typically covers wavelengths of light from 650 nm to 1000 nm.
In this range, human tissue is relatively transparent and light is able to penetrate
through biological tissue without difficulty (Ferrari & Quaresima 2012, Tuchin
2007). This range is often called an optical window, particularly due to its
characteristic transparency. Light below 600 nm is too strongly absorbed by
hemoglobin whereas light above 1000 nm is too strongly absorbed by water, both
of which limiting the penetration of light in tissue (Scholkmann et al. 2014). In
addition, the absorption spectra of deoxyhemoglobin (Hb) and oxyhemoglobin
(HbO) differ substantially in this range (Strangman et al. 2002b).
Blood oxygen variations in tissue can be determined by measuring reflected
light at two or more wavelengths recorded within the range of the optical window.
These wavelengths should be selected from both sides of the isosbestic point of the
absorption spectrum of Hb and HbO at 800 nm, where the specific extinction
coefficients of both are equal (Gibson et al. 2005). In the near-infrared (NIR)
spectral window, chromophores like Hb and HbO absorb light, and changes in their
concentration can be measured using NIRS. These temporal variations can be used
as a surrogate marker for neuronal activation, in a manner similar to the BOLD
signal used in fMRI. Actually, unlike the BOLD signal, which is principally
24
sensitive to variations in the level of Hb, the NIRS signal is able to independently
measure both HbO and Hb, thereby allowing for differentiation between changes
associated with variations in hemoglobin oxygen saturation and blood volume
(Wylie et al. 2009). Besides the most often measured Hb and HbO, NIRS can also
be used to measure the concentration of water, lipids and cytochrome in the brain
(Corlu et al. 2005, Tisdall et al. 2007, Tuchin 2007).
In NIRS devices, three different techniques are typically used related to a
specific technique of illuminating the brain (Ferrari & Quaresima 2012). In
continuous-wave (CW) technique, the illuminating light has a constant intensity
and the light attenuation is measured (Scholkmann et al. 2014). The time-domain
(TD) method uses a very short illuminating NIR pulse (a few picoseconds) and the
photon pathlength is based on time-of-flight (Torricelli et al. 2014). The frequency-
domain (FD) technique uses a light source that is modulated in intensity, and both
the attenuation and phase delay of the received light is measured (Wolf et al. 2007).
TD and FD are more sophisticated methods containing the average estimation of
the optical pathlength, which improves quantification.
2.1.2 Light propagation in human brain
The near-infrared light sources, typically laser diodes or LEDs, are placed directly
onto a subject’s scalp. In order to measure the signal from the brain, light should
penetrate relatively deep (a few centimeters) through the scalp, skull and
cerebrospinal fluid (CSF), and back again to the detector. Therefore, the detector
should be far enough from the source. The average pathway that light forms
between the source and detector is shaped like a banana (Hoshi 2003, Okada &
Delpy 2003, Okada et al. 1997) and is commonly known as banana-shaped area in
colloquial language. The penetration depth and exact optical path depend on the
optical properties of the tissue layers that are penetrated by light. Moreover, the
distance between the source and the detector has a significant influence on light
penetration. Light propagation in tissue has been studied extensively using different
mathematical models, phantom measurements and other experimental methods
(Okada & Delpy 2003, Tuchin 2007). The accuracy of these results depends on how
well the optical properties of the tissue are known (Cheong et al. 1990).
In the brain, optical penetration depth depends on the optical properties of the
scalp, skull, CSF and grey matter. McCormick and colleagues were the first ones
to prove experimentally, using living humans, that infrared light penetrates through
the intact scalp and skull and returns to the scalp for detection using a source-
25
detector distance of 2.7 cm in the first place (McCormick et al. 1992). Later, it has
been generally assumed that the penetration depth of the NIRS systems is
approximately half of the optode spacing (Gibson et al. 2005). This statement has
required a lot of theoretical work and a large number of stimulation studies (Arridge et al. 1992, Delpy et al. 1988, Strangman et al. 2002a).
Optical pathlength means the average route that light (photons) travels from
source to detector. It is clearly longer than the physical distance between the source
and the detector. Differential pathlength factor (DPF) is also typically defined as
absolute optical pathlength divided by the known distance between the optodes. It
has been shown that the DPF of the adult head is almost constant for source-
detector distances greater than 2.5 cm without considerable difference between
sexes or skin color (Duncan et al. 1995, Van der Zee et al. 1992). The mean value
of the DPF for adult forehead is between 5.93 and 7.25 (Duncan et al. 1995,
Essenpreis et al. 1993, Ferrari et al. 1993, Van der Zee et al. 1992, Zhao et al. 2002),
for the central somatosensory areas 7.5, and for occipital areas 8.75 (Zhao et al. 2002). DPF is shorter for newborn infants (Duncan et al. 1995, Van der Zee et al. 1992, Wyatt et al. 1990) and also for geriatric population (Bonnéry et al. 2012).
Moreover, Duncan et al. have provided equations for calculating DPF for use with
conventional NIRS according to the age of the subject (Duncan et al. 1996).
Equations can be found for four different wavelengths: 690 nm, 744 nm, 807 nm
and 832 nm. However, all of these different DPF values have been used in different
articles for calculations of hemoglobin concentrations from adult head.
Human adult brain is difficult tissue for measuring the NIRS signal because
light should penetrate approximately 1-2 cm. By increasing the source-detector
distance light penetrates deeper into brain tissue (Germon et al. 1999, Okada et al. 1997). On the other hand, the larger the distance, the more light is scattered, and
therefore it has also been suggested that the detector should not be placed further
away than 3 cm from the source (Jackson & Kennedy 2013, Lloyd-Fox et al. 2010).
In any case, the selection of the optimal source-detector distance depends on NIR
light intensity and wavelength, as well as the age of the subject and the head region
measured (Ferrari & Quaresima 2012). Source-detector distances vary widely,
between 2 cm and 7 cm (Hoshi 2003, Sassaroli 2006, Schelkanova & Toronov 2012,
Scholkmann et al. 2014, Villringer et al. 1993), and recently, the most commonly
used distances are usually 2–4 cm (Ferrari & Quaresima 2012, Huppert et al. 2009,
Rossi et al. 2012, Sorvoja et al. 2010). This range gives a penetration depth of about
1–2 cm, which is usually sufficient for sampling the adult cortex. However, the
detected light is significantly attenuated (Study I) and therefore advanced
26
techniques such as FD and TD techniques are required to enable brain signal
detection. An obvious limitation of NIRS is the lack of exact spatial information,
and also spatial resolution in general (Gibson et al. 2005, Maki et al. 1995). This
limitation is often solved by using NIRS simultaneously with multimodal
measurements, as will be discussed in Chapter 2.3.
2.1.3 Different wavelength combinations
The selection of wavelengths has substantial influence on the quality of the
measured NIRS data, and it also depends on the purpose of the measurement.
Typically, two or a few more wavelengths are selected, and the modified Beer-
Lambert law (MBLL) equations are evaluated at these selected wavelengths (Delpy et al. 1988, Scholkmann et al. 2014). Consequently, concentration changes of Hb,
HbO and HbT are quantified as a result. Theoretically, if only Hb and HbO are
measured, the highest signal-to-noise ratio (SNR) can be achieved by sampling at
the two optimal wavelengths as frequently as possible; this is usually the approach
in the commercial NIRS devices (Scholkmann et al. 2014). Besides wavelength
selection, increasing illumination power increases SNR but only to certain extent
(Study II). However, too powerful light may affect the tissue, for example by
heating or even burning it (Bozkurt & Onaral 2004).
In the case of two wavelength combinations, 660-760 nm and 830-850 nm are
typically selected. Yamashita and co-workers showed that the wavelength pair 664
and 830 nm gave six times more precise results on Hb and two times better on HbO
compared to 782 and 830 nm (Yamashita et al. 2001). Strangman and colleagues
used experimental studies and Monte Carlo (MC) simulations to conclude that for
measuring oxygenation changes, pairing 690 or 760 nm with 830 nm is better than
pairing 780 nm with 830 nm (Strangman et al. 2003). Uludag and colleagues used
also both theoretical and experimental methods to investigate the effect of
separability and cross talk between calculated Hb and HbO concentrations from
dual wavelength combinations and obtained results that are in agreement with the
earlier results mentioned above (Uludağ et al. 2004). Importantly, it was also shown
that non-optimal wavelengths can lead to cross-talk not only reducing the
quantitative accuracy with which the changes can be determined, but also changing
the shape of the time-course of the signal. Sato and co-workers got the best SNR
with the wavelength pair of 692 and 830 nm using different stimuli and different
locations on the head (Sato et al. 2004). Kawaguchi and colleagues got similar
results using MC simulations when they demonstrated that the wavelength pair 690
27
nm and 830 nm reduces crosstalk between HbO and Hb compared to the pair 780
nm and 830 nm (Kawaguchi et al. 2008). Earlier, the same group reported by using
MC simulations that the optimal wavelength range for pairing with 830 nm for the
dual-wavelength measurement of HbO and Hb is from 690 to 750 nm (Okui &
Okada 2005). However, also different opinions have been expressed. Funane and
colleagues demonstrated that SNR is maximal when both ends of wavelengths in
the range of 650–900nm were used (Funane et al. 2009). Correia and co-workers
concluded that the optimum wavelength pair is 704 ± 7 and 887 ± 12 nm (Correia et al. 2010).
The use of three or more wavelengths in the measurements enables more
chromophores to be studied simultaneously. However, SNR of hemoglobin
concentrations is highest, at least in theory, when only two wavelengths are taken
into account in calculations (Scholkmann et al. 2014). Three or more wavelengths
usually cause significant artifacts and therefore analyses considering only Hb and
HbO are acceptable without considering cross talk (Uludag et al. 2002, Uludağ et al. 2004, Uludag et al. 2004). That is also the case in this doctoral dissertation, and
that is why three or more wavelength combinations are not reviewed here. Multiple
wavelength combinations are reviewed for example by Scholkmann and co-
workers (Scholkmann et al. 2014).
2.1.4 Analysis of NIRS data and software tools
The method for converting raw NIRS time courses into time courses representing
temporal changes of the Hb and HbO concentration is based on MBLL and it is
described in detail, for example, in the following publications (Boas et al. 2001,
Cope et al. 1988, Cope 1991, Delpy et al. 1988). Hb and HbO are determined by
measuring the changes in optical density at two or more wavelengths and using the
known extinction coefficients. The extinction coefficients of Hb and HbO in every
wavelength of the NIR region are reported by Cope (Cope 1991). Summation of
Hb and HbO provides the concentration change in total hemoglobin (HbT), which
reflects the change in cerebral blood volume (CBV) and cerebral blood flow (CBF)
within the illuminated area (Hoshi et al. 2001). In addition, both DPF and the
distance between the source and the detector need to be determined and used in
calculations. Today, these calculations are a routine-like task, and therefore there
are software tools available for this purpose. Next, the most common software tools
are presented.
28
Hemodynamic Evoked Response (HoMer) / HoMer2
Currently, one of the most typically used NIRS data analysis tools is Matlab-based
HoMer and its further developed version, HoMer2. It is open source software which
calculates concentration changes of Hb, HbO and HbT automatically, based on
MBLL, from raw NIRS data when all the needed parameters are given (DPFs,
extinction coefficients and source-detector distances). It also provides tools for
filtering and signal processing, averaging and linear regression, and contains image
reconstruction capabilities. The basics of the program can be found in the following
publication (Huppert et al. 2009), and HoMer2 has been used, for example, in the
following publications (Chiarelli et al. 2015, Cooper et al. 2012a, Pinti et al. 2015,
Rummel et al. 2014).
NIRS-SPM
The other commonly used NIRS analysis package is Matlab-based software called
NIRS-SPM, which is used especially for statistical analysis of NIRS signals (Ye et al. 2009). Based on the general linear model (GLM), and Sun’s tube formula /
Lipschitz-Killing curvature (LKC) based expected Euler characteristics, NIRS-
SPM provides both activation maps of HbO, Hb and HbT, and super-resolution
activation localization (Li et al. 2012, Ye et al. 2009). It also includes a wavelet-
minimum description length (MDL) detrending algorithm to remove unknown
global trends due to breathing, cardiac, vaso-motion or other experimental errors
(Jang et al. 2009). In addition, NIRS-SPM could be used to estimate the cerebral
metabolic rate of oxygen (CMRO2) without hypercapnia by using simultaneous
measurements of NIRS and fMRI (Tak et al. 2010). NIRS-SPM software has been
used in NIRS analysis, for example, in the following publications (Ding et al. 2013,
Zhang et al. 2010).
NIRS Analysis Package (NAP)
The third popular NIRS analysis tool called NAP is a publicly available Matlab-
based toolbox for the analysis, visualization and anatomical registration of NIRS
data (Fekete et al. 2011a, Fekete et al. 2011b). NAP enables any combination of
physiological noise (cardiac, respiratory, blood pressure) and motion artifact
elimination from the data. Inference of activation could be done using two GLMs,
the SPM GLM (Friston et al. 1994) and a finite impulse response (FIR) model.
29
NAP also includes a stand-alone method for anatomical localization of NIRS
measurements in Montreal Neurological Institute (MNI) or any other standard
template space without magnetic resonance imaging (MRI). This enables
visualization of the results of GLM activation analyses as SPMs overlaid on the
cortical surface, both at individual and group level. The group analysis capabilities
are based on hierarchical GLM analysis (Beckmann et al. 2003). NAP can be
utilized directly, at least to Hitachi data.
nirsLAB
The most recent software called nirsLAB is a freely available Matlab-based
package for NIRS data (Xu et al. 2014). It can be utilized straightforward for
NIRScout/NIRSport, DYNOT and Biopac data by importing NIRS measurement
data to the software. Its main features include sensor registration (locations of
optodes on the scalp), data and event editing tools (measurement-timing
information), artifact correction and filtering as needed, computation of
hemoglobin parameters, data plotting, editing and topographic viewing including
block averaging and movies, anatomical structure identification, Level 1 and Level
2 GLM-based SPM computation, data export utilities and inter-subject
comparisons for hyperscanning studies (Xu et al. 2014).
2.1.5 Medical research applications
NIRS has been used in various research applications for several decades. The first
NIRS brain activation studies in human adults emerged in the beginning of the
1990s when Villringer and co-workers presented NIRS as a “new tool” for studying
hemodynamic changes in the brain during cognitive activation (Villringer et al. 1993). The typical findings of NIRS during brain activation were an increase in
HbO and a decrease in Hb, which were not related to alterations in skin blood flow.
At the same time elsewhere, Chance and his colleagues made similar conclusions
when they stated that NIRS can detect changes in low frequency components of the
light absorption signal from the human brain in response to brain function activity
(Chance et al. 1993). These initial studies were the starting point of the explosion
of NIRS brain activation studies.
At present, NIRS has become a relevant research tool in neuroscience (Myllylä et al. 2016, Obrig 2014). NIRS can be used on its own as a low-resolution
functional imaging technique because it is non-invasive, portable, cost-effective,
30
silent and easy to apply. It also has good motion tolerance and the option to co-
register other neurophysiological and behavioral data in a near natural environment
(Ferreri et al. 2014, Obrig 2014). Moreover, NIRS can be used for brain monitoring
at the bedside, at home, in remote locations, or in tasks that require motion, such as
walking (Strangman et al. 2006). The main drawbacks of NIRS are modest spatial
resolution and limited depth penetration (Strangman et al. 2006).
On the basis of neurovascular coupling, the increase in HbO and the smaller
decrease in Hb are an indirect measure of the neuronal activation and they can be
considered as robust markers of cerebral oxygenation (SaO2) changes caused by
cortical activation (Ferreri et al. 2014, Steinbrink et al. 2006). These are the basic
features of NIRS measurements, and they can be seen when using different tasks
such as visual tasks where the subject is watching some visual stimulation (Obrig et al. 2000, Villringer et al. 1993, Wolf et al. 2002, Wylie et al. 2009), basic motor
tasks where the subject is tapping her/his finger or toe (Bajaj et al. 2014, Obrig et al. 1996, Strangman et al. 2006), and different cognitive tasks such as working
memory tasks (Hoshi 2003, Ogawa et al. 2014), simple mental arithmetic tasks
(Pfurtscheller et al. 2010a, Villringer et al. 1993) or language processing tasks
(Rossi et al. 2012).
NIRS has also been used in both breath holding (BH) (Molinari et al. 2006,
Schelkanova & Toronov 2012) and carbon dioxide (CO2) inhalation tasks (Selb et al. 2014, Virtanen et al. 2009), as well as in hyperventilation (HV) tasks (Holper et al. 2014, Virtanen et al. 2009, Watanabe et al. 2003, Yang et al. 2015). These
measurements are not so unequivocal and the reported results differ. The main
reason for this is that both hyper- and hypocapnia seem to be very subject-
dependent. On average, HbO increases and Hb remains stable or decreases during
hypercapnia (Molinari et al. 2006, Selb et al. 2014, Virtanen et al. 2009). On the
other hand, at the very end of the BH task Hb seems to increase, which is in line
with our results (Molinari et al. 2006). In general, HbO decreases and Hb remains
stable or increases during HV (Virtanen et al. 2009, Watanabe et al. 2003, Yang et al. 2015).
NIRS has also been used in resting state (RS) to study spontaneous
hemodynamic fluctuations and functional connectivity using the frequency range
0.009–0.1 Hz (Medvedev 2014, Obrig et al. 2000, Sasai et al. 2011). These
fluctuations are believed to reflect some types of neuronal signaling, systemic
hemodynamics and/or cerebral autoregulation (CA) processes (Tong & Frederick
2010). New, multi-channel and high-density NIRS devices have made it possible
to measure dynamic interactions between brain areas through temporal correlations
31
of NIRS signals, and therefore to derive NIRS-based ‘functional connectivity’
similar to the functional connectivity measured by the fMRI BOLD signal (Auer
2008, Biswal et al. 1995, Medvedev 2014). Using fNIRS, RSFC has been found
from the primary sensory system (Lu et al. 2010, White et al. 2009), motor system
(Lu et al. 2010, White et al. 2009), high-level functional systems like language
system (Zhang et al. 2010) and from the head as a whole (Mesquita et al. 2010).
The advantages of the NIRS are excellent temporal resolution, affordability, silence,
portability and the ability to simultaneously measure both HbO and Hb.
In addition, thanks to NIRS’ portability and relative motion robustness it can
be used for brain monitoring during both walking and running (Atsumori et al. 2010,
Miyai et al. 2001, Suzuki et al. 2004, Suzuki et al. 2008), riding a bicycle inside or
even outside (Piper et al. 2014) and driving a car (Yoshino et al. 2013a, Yoshino et al. 2013b). This makes a big difference compared to fMRI, for example.
During the last decade, NIRS has also been used in brain computer interface
(BCI) studies (Coyle et al. 2007, Sitaram et al. 2007) in which the subject performs
tasks known to affect brain oxygenation; the use of NIRS allows the subject to
communicate with a computer or other external device, such as a robot hand. Very
recently, three different mental activities were used in fNIRS-BCI. Hong and co-
workers used mental arithmetic, right-hand motor imagery and left hand imagery
tasks whereas Power and colleagues used mental arithmetic task, mental singing
task and unconstrained, “no-control” state to investigate the feasibility of a three-
class fNIRS-BCI system (Hong et al. 2015, Power et al. 2012). Both of these
studies pointed out the potential of a three-state fNIRS-BCI system, and perhaps
fNIRS-BCI can be used as a clinical BCI system in the future.
2.1.6 Potential clinical applications
Although NIRS has become a relevant research tool in neuroscience it is still
largely unknown to clinical neurologists (Obrig 2014). One reason for this might
be that even though all NIRS monitors are based on the same basic principles, the
devices differ in their methods to generate data and in their algorithms to process
it, which makes comparisons between studies difficult, and may have prevented
NIRS from becoming a clinical tool (Bevan 2015). However, this is changing little
by little and new approaches are increasingly being presented. In this chapter, some
brain disorders and medical procedures are presented where NIRS has shown to
have potential to become a clinical tool and where it has been tested.
32
Oximeter
Perhaps most typically, NIRS has been used as a bed-side oximeter to monitor SaO2
during cardiac and vascular procedures like surgery where there is some evidence
that NIRS-guided brain protection protocols might lead to a reduction in
perioperative neurological complications (Moerman et al. 2015, Murkin & Arango
2009, Obrig 2014, Smith 2011, Zheng et al. 2013). For example, a significant
reduction in perioperative stroke rate was observed in patients in whom cerebral
oximetry was used to optimize and maintain intraoperative SaO2 (Murkin &
Arango 2009).
Cerebrovascular disease
Cerebrovascular diseases such as stroke are also very widely studied field of NIRS
clinical applications. Stroke may be considered as a chronic with three stages:
prevention, (sub)acute and rehabilitation, all of which have been experimentally
studied by NIRS (Obrig & Steinbrink 2011). When a diagnosis of “stroke” is
clinically suspected it is mandatory to differentiate between ischemia and
intracerebral hemorrhage; this requires imaging, usually computed tomography
(CT) or MRI (Obrig 2014). NIRS cannot reliably differentiate between the two, but
during the subacute stage after diagnosis and initial treatment, continuous
monitoring of cerebrovascular parameters can be deemed the most promising
potential field for NIRS in clinical neurology (Obrig 2014). Budohoski and co-
workers demonstrated that impaired autoregulation in the first 5 days after stroke
is predictive of delayed cerebral ischemia and that autoregulation can be reliably
determined by NIRS (Budohoski et al. 2012). In addition and importantly, they
showed that NIRS demonstrated dysautoregulation on average 1 day earlier than
transcranial Doppler. Aries and his colleagues showed in a pilot study that 96% of
systemic desaturations were rapidly followed by local cerebral desaturations
measured by NIRS and that local nocturnal cerebral desaturations were more than
twice as likely in the affected hemisphere (Aries et al. 2012). Pizza and co-workers
documented asymmetrical nocturnal cerebral hemodynamic changes induced by
sleep-disordered breathing during the acute/subacute phase of stroke using NIRS
(Pizza et al. 2012). However, it needs to be noted that all of the above-mentioned
results require confirmation in a larger sample size. There is also support for the
potential of NIRS in risk assessment during the primary and secondary prevention
33
stages of cerebrovascular disease (Obrig 2014, Obrig & Steinbrink 2011, Oldag et al. 2012, Vasdekis et al. 2012).
Epileptic disorder
Another very widely studied field of clinical NIRS applications is epileptic disorder
(Obrig 2014). It is somewhat difficult to draw simple conclusions about the matter
because it seems that the oxygenation patterns measured by NIRS are more or less
specific to seizure type. Already the first reported results were contradictory since
Villringer and co-workers suggested that complex partial seizures (CPS) induce
large increases in HbT over the frontal lobe (Villringer et al. 1994) whereas
Steinhoff and his colleagues reported a decrease in regional oxygen saturation over
the frontal lobe ipsilateral to the focus (Steinhoff et al. 1996). Only a few years
later, Watanabe and co-workers used multichannel NIRS and reported an increase
in regional CBV (rCBV) in the region of the focus in all twelve cases (Watanabe et al. 1998). Another two years later, Sokol and co-workers studied differences in
SaO2 for CPS and rapidly secondarily generalized CPS (RCPS) and found that it
increased for CPS and decreased for RCPS, suggesting that NIRS is able to
distinguish these patterns from each other (Sokol et al. 2000). Haginoya and his
colleagues indicated new evidence of NIRS’ potential to distinguish between
different epileptic seizures when they studied 15 children having various types of
pediatric epileptic seizures (Haginoya et al. 2002). In that study, they showed that
convulsive seizures were associated with a remarkable increase in CBV, absence
seizures were associated with a mild decrease or no change in CBV of the frontal
cortex, an initial transient decrease in CBV was observed in some types of
convulsive seizures, an ictal increase in CBV changed to an ictal decrease in the
course of tonic epilepticus and there was definite heterogeneity in the CBV changes
during tonic spasms in patients with West syndrome. On the other hand, Buchheim
and her colleagues showed a reproducible decrease in HbO and an increase in Hb
during absence seizures of adults indicating a reduction of cortical activity
(Buchheim et al. 2004). Watanabe and co-workers proved NIRS’s potential in focus
localization of pharmacologically intractable epilepsy when NIRS identified the
affected hemisphere in 28 of a total of 29 cases (Watanabe et al. 2002). In addition,
Slone and his colleagues found hemodynamic changes in the frontal lobe up to 15
minutes before the onset of temporal lobe seizures, indicating that NIRS may be
useful for the prediction of seizure onset in subjects with temporal lobe epilepsy
(Slone et al. 2012). In summary, it can be noted that NIRS has proved its potential
34
in both focus localization and monitoring oxygenation responses during epileptic
activity.
Other potential NIRS studies
NIRS has also been used, for example, in studying Alzheimer’s disease (Arai et al. 2006), migraine (Liboni et al. 2007), traumatic brain injury (Irani et al. 2007) and
in case studies of sleep (Näsi et al. 2011, Virtanen et al. 2012). Even though there
are multiple avenues for NIRS to become a successful clinical monitoring tool for
brain disorders its status has still remained preliminary for some reason.
2.2 Physiological brain measurements
Traditionally, it was thought that physiological signals such as cardiorespiratory
pulsations are noise that masks the most important focus of neuroimaging, namely
the neuronal signal. Today, with the invention of glymphatic brain clearance
pulsations, it has been shown that every form of physiological pulsation, such as
vasomotor waves, respiration and cardiovascular pulses has a strong influence on
brain physiology and well-being. Thanks to this advance in the understanding of
brain physiology, all of the physiological pulsations are regarded as a source of
information rather than noise on brain status in this doctoral thesis.
2.2.1 Functional magnetic resonance imaging
MRI was discovered already in the 1970s by Paul Lauterbur and Peter Mansfield
(Lauterbur 1973, Mansfield 1977) whose research work was rewarded in 2003 by
the Noble prize of medicine. MRI enabled the discovery of fMRI, which allows the
monitoring of systemic brain network activity and neurovascular coupling.
Conventional BOLD fMRI
The BOLD contrast, discovered by Ogawa in 1990, arises from the magnetic
properties of hemoglobin. To be precise, blood has different magnetic susceptibility
based on whether it is oxygenated or deoxygenated. Paramagnetic
deoxyhemoglobin in the capillary and venous blood alters the regional magnetic
susceptibility and acts as a contrast agent for fMRI (Ogawa et al. 1990, Ogawa et al. 1992). Fox & Raichle showed already in 1986 with positron emission
35
tomography (PET) that in normoxia, an increase in the brain’s neuronal activity
leads to a higher metabolic demand and for that reason, the brain becomes over-
oxygenated, which leads to an increase in the CBV (Fox & Raichle 1986). This
functional hyperemia decreases the proportion of deoxyhemoglobin and T2* values,
which increases fMRI BOLD signal. This rise is known as the BOLD response and
it can be detected in fMRI after 3-5 s delay to neural activity. In general, this
phenomenon is called neurovascular coupling.
Between 1990 and 2005, the focus of fMRI was mainly on detecting task
activation-related hemodynamic responses with increasing accuracy. In 1995,
Biswal and his colleagues employed a new analytic approach to investigating
spontaneous fluctuations when subjects were just resting (Biswal et al. 1995). In
that study, they were able to shown slow but repeatable bilaterally synchronized
low frequency oscillations (LFOs) in the cortical motor regions at low frequencies
(<0.1 Hz). That was the starting point for a new subfield of the fMRI research, the
field of resting state fMRI (RS-fMRI); it took approximately 10 years before the
focus in neuroimaging shifted exponentially towards RS-fMRI (Flodin 2015).
Today, it is the most often used method to analyze brain function (Friston 2011).
RS BOLD signal LFOs have been linked to functional connectivity of brain
networks and vasomotor waves (Biswal et al. 1995, Kiviniemi et al. 2000, Majeed et al. 2011b, Thompson et al. 2014).
The BOLD signal is influenced not only by neuronal activity, but also by the
effects of cardiovascular and respiratory processes (Birn et al. 2006, Birn et al. 2008, Chang et al. 2009, Chang & Glover 2009). These physiological noises
(respiration ~ 0.3 Hz and heart rate (HR) ~ 1 Hz) obscure neuronal signals because
they unavoidably alias the classical BOLD signal sampled at ~0.5 Hz. This makes
definite separation of fluctuation sources a highly demanding task from a single
scan (Kiviniemi et al. 2005). In addition to aliasing, the noise and signals can
overlap in time, be anatomically co-localized and have common temporal
frequencies (Chang & Glover 2009). In conclusion, the imaging speed of
conventional BOLD fMRI is reasonably well-suited to the hemodynamic response
time used for observation of the BOLD contrast mechanism but it fails to observe
physiological events (especially cardiac) occurring over a shorter time scale
(Hennig et al. 2007). This is the main reason why faster fMRI imaging sequences
are needed and why they have recently been developed.
36
MREG and other ultrafast imaging sequences
Ultra-fast MRI techniques allow the observation of functional signal in the brain
20-40 times faster than using conventional fMRI BOLD. Recent ultrahigh temporal
resolution fMRI imaging techniques are inverse imaging (InI) (Lin et al. 2012),
MREG (Assländer et al. 2013), generalized inverse imaging (GIN) (Boyacioglu &
Barth 2013) and multi-slab echo-volumar imaging (MEVI) (Posse et al. 2013), all
of which enable critical sampling of brain data without spurious aliasing. This
means that all physiological noise (especially respiration and cardiac) can be
reliably filtered away from the data, which improves the accuracy of the fMRI data,
such as LFOs. Furthermore, the faster scanning increases the statistical power of
the analysis thanks to a marked increase in the number of samples per scan. As a
result of increased temporal resolution, it is possible to map novel physiological
phenomena, such as cardiorespiratory glymphatic pulsations (Kiviniemi et al. 2016). Combining good spatial and temporal resolution is a challenge in ultra-fast
MRI; the faster one gets, the less spatially accurate information one gets. However,
100 ms ultra-fast fMRI signal can have a nominal spatial resolution of 3–4 mm,
which is comparable to most conventional BOLD fMRI data (Assländer et al. 2013).
Analysis of fMRI data
Independent component analysis (ICA) is a data-driven method which can
effectively detect and separate multiple functional units known as resting state
networks (RSNs) in different parts of the brain from fMRI data (Beckmann et al. 2005, Kiviniemi et al. 2003, Kiviniemi et al. 2009, Smith et al. 2009). One of the
most prominent RSNs is the so-called default mode network (DMN) that is
activated during the RS and presents a deactivation pattern in task-fMRI studies
(Greicius et al. 2003, Raichle et al. 2001). DMN can typically be divided into two
smaller subnetworks called the ventromedial DMN (DMNvmpf) and posterior
cingulate cortex DMN (DMNpcc). Another typically used method for identifying
RSNs from fMRI data is seed-based correlation analysis (Biswal et al. 1995, Cole et al. 2010, Van Dijk et al. 2010). Seed-based correlation analysis reveals also the
functional connectivity of networks. Simultaneity of different RSNs reflects the
state of neurophysiology (Biswal et al. 2010, Boly et al. 2011).
37
Software tools of fMRI
Perhaps the most common fMRI software tool is FMRIB software library (FSL),
and that is why fMRI data (both conventional and ultra-fast) are often processed
and analyzed using the FSL pipeline (Jenkinson et al. 2012). One of the first steps
is usually removing the first time points from the beginning of the data to minimize
T1-relaxation effects. Head motion can be corrected with FSL MCFLIRT software
(Jenkinson et al. 2002) and brain extraction can be done using FSL BET software
(Smith 2002). ICA is also part of the FSL pipeline and it can be found from FSL
multivariate exploratory linear optimized decomposition into independent
components (MELODIC) part (Beckmann & Smith 2004).
Other widely used fMRI software tools are AFNI (Cox 1996, Cox 2012) and
SPM8 (Ashburner et al. 2012). It needs to be highlighted that AFNI was the first
software tool for analyzing RS data.
2.2.2 Blood pressure measurements
Arterial pressure (AP) oscillations, better known as vasomotor waves or Mayer
waves, occur spontaneously in conscious subjects at a frequency lower than
respiration (~0.1 Hz in humans) (Julien 2006, Mayer 1876). Two theories have been
proposed to explain the generation of Mayer waves: the pacemaker theory and the
baroreflex theory. The pacemaker theory is based on the central oscillations of
sympathetic activity of an endogenous oscillator located either in the brainstem or
in the spinal cord (Julien 2006, Kiviniemi et al. 2010). Because Mayer waves are
enhanced during states of sympathetic activation, they are proposed to be an
indirect measure of efferent sympathetic nervous activity (SNA) (Julien 2006). The
baroreflex theory suggests that the generation of slow oscillations is a resonance
phenomenon of the baroreflex loop (Julien 2006, Pfurtscheller et al. 2011).
However, there is still no clear knowledge of the underlying physiology of Mayer
waves. Quite recently, slow blood pressure oscillations have also been related to
self-based, voluntary movements (Pfurtscheller et al. 2010b, Pfurtscheller et al. 2012a).
For instance, by measuring the speed of arterial pulse wave velocity (PWV) or
pulse transit time (PTT) from the blood vessels as a continuous measure, one is
able to noninvasively estimate the arterial blood pressure (BP) because both PTT
and PWV have been shown to have good correlation with systolic blood pressure
(Gesche et al. 2012, Gribbin et al. 1976, Sorvoja 2006). PTT can be determined,
38
for example in MRI, using fiber optic-based sensors (Myllylä et al. 2012, Myllylä et al. 2011). PTT is commonly measured between the R-spike of the
electrocardiograph (ECG) and photoplethysmograph (PPG) curve of finger pulse-
oximetry (Gesche et al. 2012, Myllylä et al. 2013a).
2.2.3 Electroencephalography
Berger first described the rhythmic behavior of the human brain activity during RS
(Berger 1929). EEG is a medical imaging technique that measures the spontaneous
electrical activity of the brain from the scalp over a period of time by multiple
electrodes. The voltage fluctuations on the scalp are presumed to result from the
synchronized synaptic activity of a large number of neurons (Moosmann et al. 2003,
Niedermeyer & da Silva 2005). Because EEG is typically measured from the scalp
through the skull, which has high resistivity to electric field, the measured voltages
are small (~10-100 µV). As a result, EEG has poor spatial resolution but it still
offers very good temporal resolution.
During ongoing brain activity, EEG is dominated by spontaneously occurring,
spatially distributed, oscillatory rhythms in characteristic frequency bands
(Moosmann et al. 2003). These frequency bands can be classified, for example, as
follows: delta < 4 Hz, theta 4-8 Hz, alpha 8-13 Hz, beta 13-30 Hz and gamma > 30
Hz (Kropotov 2010). The focus of interest of our research are the slow delta waves
and more precisely, very low frequency (VLF) oscillations (0.009-0.08 Hz), which
can be measured using direct current EEG (DC-EEG).
Such VLF oscillations, up to 1-2 mV electrical potential of mammalian brain
tissue, were observed for the first time in the 1950s in electro-cortical experiments
on rabbits (Aladjalova 1957). In the 1970s, large amplitude brain-potential shifts
evoked by respiratory acidosis in animal experiments were suggested to originate
from a potential difference across the blood-brain barrier (BBB) (Revest et al. 1993,
Revest et al. 1994, Woody et al. 1970). Comparable millivolt-scale direct current
(DC) shifts are seen upon HV or hypoventilation in the human scalp DC-EEG
(Voipio et al. 2003). Nita and co-workers observed even larger DC-EEG shifts from
cats during hyper- and hypoventilation (Nita et al. 2004). VLF oscillations in the
human DC-EEG are synchronized with faster cortical EEG oscillations and they
are phase-locked with slow fluctuations in brain excitability (Hiltunen et al. 2014,
Monto et al. 2008, Vanhatalo et al. 2004), suggesting a link between VLF
oscillations and the mechanisms of neurovascular coupling at the level of BBB
(Vanhatalo et al. 2004).
39
2.2.4 Physiological signals offered by anesthesia monitoring
Human physiological signals affect and counteract each other, and that is why it is
very important to measure them as exactly as possible. Two important
cardiorespiratory signals are end-tidal carbon dioxide (EtCO2) and partial pressure
of carbon dioxide in the arterial blood (PaCO2), both of which play an important
role in CA: an increase induces cerebral vasodilation, thereby increasing CBF,
while a decrease induces cerebral vasoconstriction, thereby decreasing CBF
(Bhambhani et al. 2007, Scholkmann et al. 2013). To maintain stable and adequate
nutritional CBF, the brain’s vasculature must respond to changes in arterial BP or
intracranial pressure (Claassen et al. 2009, Van Beek et al. 2008). Also HR is
affected by normal breathing due to coupling and interaction between the cardiac
and respiratory systems (Indic et al. 2008), and neural events locked to heart beats
before stimulus onset predict the detection of a faint visual grating in two
multifunctional RSNs (Park et al. 2014). HR variability (HRV), which can be
defined as changes in the beat-to-beat interval or in the instantaneous HR, has been
shown to have connection with the RSN dynamics (Chang et al. 2013). Its high
frequency component in the range of 0.15–0.4 Hz is attributed to respiration-
induced HR modulation while its lower frequency component in the range of 0.05–
0.15 is not fully understood. Another similar kind of signal is respiratory volume
per time (RVT), which means breath-to-breath metric of respiratory variation,
computed from pneumatic belt measurements of chest expansion (Chang & Glover
2009). RVT is believed to capture breathing-induced changes in arterial CO2.
Because of these huge complexities, synchronous multimodal measurement is the
only way to obtain precise assessment of these complex dynamics and the
interactions between physiological variables and brain activity.
In fact, most of the physiological phenomena mentioned above can be
measured from humans by using an anesthesia monitor. Respiration (EtCO2) is
measured from the mouth or nose with an oxygen mask, and cardiac-related signals
are measured from the chest with ECG or from a finger with a PPG sensor. The
PPG sensor also indicates the level of peripheral capillary oxygen saturation (SpO2).
BP can be measured from the arm by using a cuff-based non-invasive method or
invasively straight from the artery. However, NIBP provided by the anesthesia
monitor can give only one diastolic and systolic blood pressure value for each
measurement, at most at 1-2-minute intervals. That is why it can be used only for
slow trend monitoring. All of these signals can be used for real-time patient
monitoring.
40
2.3 NIRS in multimodal brain imaging combinations
Currently, there is an increasing trend in neuroimaging to utilize simultaneous
multimodal measurements because investigating the functional principles of the
human brain requires them (Myllylä 2014). No single modality is able to provide
both good spatial and temporal resolution of human brain activity. For example,
despite the wide use of fMRI, the neurophysiological sources of spontaneous brain
activity fluctuations are still relatively unknown. It needs to be noted that fMRI
relies on an indirect signal, the BOLD contrast, which is caused by an increase in
oxygen delivery, strongly exceeding the metabolic demand (Steinbrink et al. 2006).
2.3.1 NIRS vs. fMRI
Combined NIRS and fMRI offers both good temporal (NIRS) and spatial (fMRI)
resolution in detecting cerebral blood oxygenation changes during brain activation.
FMRI can be used to detect global BOLD response based on blood paramagnetic
properties and NIRS signal can be used to detect the same phenomena locally based
on blood optical properties after neuronal activity (Myllylä et al. 2016). NIRS
measurements have lower spatial resolution and depth penetration than fMRI, but
on the other hand, they offer better specificity due to blood optical properties.
Accordingly, several oxygenation-related parameters like Hb, HbO and HbT can
be determined from the NIRS data, as opposed to fMRI. NIRS also offers excellent
temporal resolution compared to fMRI. Both of these advantages can be used to
gain a better understanding of the nature of the physiological rhythms, the
hemodynamic response to neuronal activation and the origin of the BOLD signal
(Strangman et al. 2002b).
The first simultaneous NIRS and fMRI measurement was carried out in 1996
to measure cerebral blood oxygenation changes during human brain functional
activation (Kleinschmidt et al. 1996). The main purpose was to validate both
methods using the finger tapping method, which was known to increase MRI signal
intensity in motor cortex due to Hb decrease. About one year later, a simultaneous
NIRS-PET study confirmed the validity of the NIRS measurements in human adults
as a statistically significant correlation was found between the changes in CBF by
PET and changes in HbT measured by NIRS during rest and during performance
of a calculation task and Stroop task (Villringer et al. 1997). Toronov and co-
workers have shown qualitative spatial and temporal correspondence between
41
BOLD fMRI and NIRS signals using both amplitude and phase measures (Toronov et al. 2001a, Toronov et al. 2001b).
Since the turn of the 21st century, the amount of simultaneous NIRS and fMRI
studies has increased significantly (Steinbrink et al. 2006). Most of the studies deal
with different tasks such as motor (Gagnon et al. 2012, Gratton et al. 2005, Huppert et al. 2006, Tak et al. 2010), visual (Schroeter et al. 2006, Toronov et al. 2007),
hypercapnia (BH or CO2 inhalation) (Alderliesten et al. 2014, MacIntosh et al. 2003, Tong et al. 2011, Yücel et al. 2014) and working memory tasks (Cui et al. 2011, Sato et al. 2013). In task-based activation studies, HbO typically increases
and Hb decreases due to neuronal activation. However, Fujiwara and co-workers
showed that in the case of brain tumor, both HbO and Hb increased on the lesion
side during a contralateral hand grasping task, indicating the occurrence of rCBF
increases in response to neuronal activation (Fujiwara et al. 2004). At the same
time, BOLD-fMRI revealed only a small activation area or no activation on that
side. In addition, intraoperative brain mapping identified the primary sensorimotor
cortex that was not demonstrated by BOLD-fMRI. The false-negative activations
were associated with an increase in Hb during activation, which may have been
caused by the atypical evoked SaO2 changes. On the non-lesion side, NIRS
demonstrated a typical decrease in Hb and an increase in HbO, and BOLD-fMRI
demonstrated clear activation areas in the primary sensorimotor cortex. Very
recently, also negative BOLD responses to visual stimulation were studied by NIRS
(Maggioni et al. 2015). In that article, it was confirmed that HbO and Hb changes
were the key elements in the investigation of the hemodynamic control mechanisms
underlying negative BOLD responses.
Very recently, some simultaneous RS NIRS-fMRI studies have also been done
(Cooper et al. 2012b, Duan et al. 2012, Sasai et al. 2012, Tong & Frederick 2010,
Toronov et al. 2013). These studies can be divided into three fields where the first
investigates LFOs using correlation analysis between NIRS and fMRI time series
and attempts to shed light on the question of what forms the BOLD signal (Sasai et al. 2012, Tong & Frederick 2010), the second field focuses on RSFC and on trying
to find similar networks (Duan et al. 2012), and the third uses NIRS signals as
regressors due to minimizing the background physiological noise from the fMRI
BOLD signal (Cooper et al. 2012b). In RS, especially LFOs are the main focus of
interest because they are typically observed in blood-related brain functional
measurements and their physiological origin and implications are not yet fully
understood (Tong & Frederick 2010). Tong and co-workers have shown using
simultaneous NIRS and fMRI measurements that a major component of the LFOs
42
arises from fluctuations in the blood flow and hemoglobin oxygenation at a global
circulatory system level (Tong & Frederick 2010), but not all the RSNs identified
with ICA are affected equally by the systemic LFOs (Tong et al. 2013). Some of
the RSNs were strongly correlated with the peripheral signals measured by NIRS,
while, for example, DMN was not. However, it needs to be noted that they used
conventional BOLD with slow TR, which surely has an impact. Duan and co-
workers have demonstrated that multichannel fNIRS is capable of providing
comparable measures of RSFC to fMRI (Duan et al. 2012). Interestingly, in that
study HbO demonstrated higher RSFC similarity with BOLD than Hb. Cooper and
his colleagues have shown that simultaneous NIRS recordings have significant
benefit to the regression of low-frequency physiological noise from fMRI data
(Cooper et al. 2012b).
One good example of the clinical potential of simultaneous NIRS-fMRI
measurements is the ability to reveal various hemodynamic and metabolic changes
between subcortical vascular dementia (SVD) patients and healthy controls, which
may be used for early detection or monitoring (Tak et al. 2011). In that study, Tak
and co-workers showed a robust reduction in the ratio of oxygen supply to its
utilization, which implies a loss of metabolic reserve.
2.3.2 NIRS vs. EEG
NIRS can be used to measure SaO2 and hemodynamics, whereas EEG measures
the electrophysiological signals from neurons. While EEG is very powerful in
detecting brief processes in the range of 100 ms, NIRS provides integration over a
longer timeframe as well as better localization (Wallois et al. 2012). Simultaneous
NIRS-EEG measurements are complementary as they give information about the
relationship between fluctuations in the cerebral hemoglobin oxygenation state and
neuronal activity with high temporal resolution (Hoshi et al. 1998). Therefore,
NIRS-EEG is a promising technique for the non-invasive study of neurovascular
coupling, and NIRS and EEG signals have shown significant correlations across
subjects (Chen et al. 2015). However, it has been exploited quite rarely, possibly
because these methods have low spatial resolution and they are therefore more
commonly combined with fMRI. Nevertheless, combined NIRS-EEG studies have
become more common today, especially in the fields of epilepsy and brain
computer interface.
One of the first simultaneous NIRS-EEG studies was carried out by Moosmann
and co-workers in 2003 when they studied cross-correlation of alpha rhythms
43
between NIRS and EEG. They found a positive cross-correlation in the occipital
cortex between alpha activity and concentration change of Hb, which peaked at a
relative shift of about 8 s. Based on that, they suggested that alpha activity in the
occipital cortex is associated with metabolic deactivation (Moosmann et al. 2003).
On the other hand, Pfurtscheller and co-workers showed that the positive LFO
peaks of HbO preceded the central EEG beta (alpha) power peak by 3.6 ± 0.9 s
during RS and motor act, suggesting that it can be indicative of a slow excitability
change of central motor cortex neurons (Pfurtscheller et al. 2012a, Pfurtscheller et al. 2012b). Koch and colleagues investigated also alpha rhythms using NIRS and
EEG during rest and visual stimulation. Their most important finding was that
individual alpha-frequency (IAF) correlates with the oxygenation response to
visual stimulation: A high IAF predicts a low alpha-amplitude at rest, small
electrophysiological and vascular responses to stimulus and vice versa (Koch et al. 2008). Based on these findings, they assumed that the relationship between IAF
and both neuronal and vascular response stems from the size of the network
recruited for visual processing.
In different epilepsy studies, combined NIRS-EEG is still a relatively new non-
invasive neuroimaging technique with potential for long-term monitoring of the
epileptic brain (Peng et al. 2014). Simultaneous recordings of EEG and NIRS allow
the investigation of the hemodynamic changes associated with epileptiform events
before, during and after interictal epileptiform discharges (IEDs) and seizures
detected on scalp EEG (Machado et al. 2011, Pouliot et al. 2014). Spatially
concordant increase in rCBV and HbO at the epileptogenic focus has been found
with focal and absence epilepsy using NIRS-EEG (Machado et al. 2011, Pouliot et al. 2012, Roche‐Labarbe et al. 2008). Contrary to that, Furusho and co-workers
observed an increase in Hb and a slight decrease in HbO and rCBV in a boy with
temporal lobe epilepsy (Furusho et al. 2002). However, NIRS-EEG is easy to use,
noninvasive, and safe, and offers the best available combination of spatial and
temporal resolutions (Roche‐Labarbe et al. 2008). In summary, simultaneous
NIRS-EEG has already shown its potential in long-term recording, seizure
detection, contributing favorably to pre-surgical investigation, localization of the
epileptic focus and characterization of the complex local and remote oxygenation
changes occurring during temporal lobe seizures, frontal lobe seizures and interictal
focal spikes (Furusho et al. 2002, Gallagher et al. 2008, Nguyen et al. 2012,
Nguyen et al. 2013, Peng et al. 2014, Pouliot et al. 2014).
BCI research is advancing very rapidly thanks to multimodal imaging systems
called “hybrid BCIs” (Pfurtscheller et al. 2010c) that allow users to control external
44
devices or computers by their intentions decoded from their brain activity (Lee et al. 2015). One good example is hybrid NIRS-EEG BCI, which enables generating
more control commands with better accuracy than in the case of a single modality
(Khan et al. 2014). Hybrid NIRS-EEG BCI enhances both classification and
performance for motor execution and steady-state visual evoked potentials (Fazli et al. 2012, Tomita et al. 2014). Hybrid NIRS-EEG BCI can be used as a practical
self-based motor imagery system as well (Koo et al. 2015).
Combined NIRS-EEG has also been used to study language-related, neural and
cerebral functional activity, which is reviewed by Wallois and colleagues in 2012
(Wallois et al. 2012). Furthermore, LFOs have been studied between simultaneous
measurement of prefrontal NIRS and EEG (Pfurtscheller et al. 2012a, Pfurtscheller et al. 2012b). One new direction in this field is also the development of a wireless
portable system for functional brain imaging (Lareau et al. 2011, Sawan et al. 2013).
Finally, I would like to mention the simultaneous epidural NIRS and cortical
electrophysiology recording tool, which has been shown to be a promising tool for
studying local hemodynamic signals and neurovascular coupling (Zaidi et al. 2015).
At least in that case, one can be absolutely sure that NIRS is measuring the signal
from the brain and is not affected by the hemoglobin concentration changes
occurring in the scalp tissue.
2.3.3 NIRS vs. NIBP
LFOs around 0.01-0.1 Hz have been studied using NIRS and NIBP to get more
information about the interaction between the brain and the heart. Pfurtscheller and
his colleagues studied the phase coupling between slow fluctuations in
cardiovascular and central autonomic networks by analyzing BP time series and
NIRS time series (Hb, HbO) at optodes placed over the prefrontal cortex
(Pfurtscheller et al. 2011). They got two interesting results: first, clear BP-coupled
HbO oscillations around 0.1 Hz were observed in about 60% of the subjects and
second, the phase shifts between slow arterial BP, HR beat-to-beat intervals and
HbO oscillations were relatively stable, subject-specific and similar during rest and
movement. Rowley and co-workers reported also a close relationship between
LFOs in cerebral HbO and in systemic mean arterial pressure (MAP) investigated
by wavelet cross-correlation between the two time series (Rowley et al. 2006).
Moreover, Kirilina and her co-workers studied physiological noises of fNIRS from
the prefrontal cortex and found that NIRS is highly correlated with time-lagged
MAP and HR fluctuations with a period of about 10 seconds, often referred to as
45
Mayer waves (Kirilina et al. 2013). They showed also that these signals can be used
to develop physiological regressors which can be used for physiological de-noising
of fNIRS signals.
On the other hand, Minati and his colleagues proved the necessity of
simultaneous, continuous BP measurements alongside NIRS by manipulating BP
actively (Minati et al. 2011). When BP was actively manipulated, NIRS was not
able to track differences in visual stimulus duration but showed a very large
pressure-related response. Without manipulation, NIRS was able to track those
differences, which demonstrates that BP fluctuations can exert confounding effects
on brain NIRS through expression in extracranial tissues and within the brain itself.
Very recently, it has also been shown that VLF oscillations (0.02-0.07 Hz) of
NIRS and BP are affected by aging and cognitive load (Vermeij et al. 2014). Their
results suggested that not only local vasoregulatory processes, but also systemic
processes influence the cerebral hemodynamic signals and due to that, they
concluded that the effects of age and BP should be taken into account in the analysis
and interpretation of neuroimaging data that rely on blood oxygen levels.
2.3.4 NIRS/BOLD vs. Cardiorespiratory signals
LFOs have been found to correlate between different cardiorespiratory signals and
both NIRS and fMRI. Wise and co-workers found significant correlation between
the fluctuations in EtCO2 measured by a capnograph and BOLD fMRI signal at a
low frequency at RS (Wise et al. 2004). On the other hand, Birn and his colleagues
showed significant time-lagged correlations between RVT and fMRI time series
(Birn et al. 2006).
Schmueli and co-workers showed significant correlations between the cardiac
rate and BOLD signal time courses, particularly negative ones in the grey matter at
time shifts of 6-12 seconds and positive correlations at time shifts of 30-42 seconds
(Shmueli et al. 2007). Also others have reported the influence of cardiac activity
on the fMRI signal (Bhattacharyya & Lowe 2004, Dagli et al. 1999). However, it
needs to be noted that the TR has been relatively slow in all of these measurements,
lower than Nyquist frequency. Thus, aliasing must be an important issue and proper
attention should be paid to that aspect.
Tong and his colleagues have brought cardiac LFO analysis one step forward
because they have studied correlations between HbT from the fingertip and BOLD
fMRI signal changes in RS ICs (Tong et al. 2013). In that study they found that not
all the RSNs identified with ICA were equally affected by the systemic LFOs; some
46
of them were strongly correlated with the peripheral signals. The RS BOLD signals
in brain regions of sensorimotor, auditory and visual cortices were significantly
correlated with systemic LFOs, whereas, for example, DMN was not (Tong et al. 2013).
Nevertheless, because of both respiration and HR, it is important to regress out
physiological noise, especially from the fMRI signal. For that reason it is important
to measure them as accurately as possible and preferably utilize multimodal
imaging.
47
3 Aims of the study
The purpose of this study was to integrate and synchronize NIRS with ultra-fast
multimodal neuroimaging. The particular aims of the study were:
1. to study NIR light propagation into the human cerebral cortex
2. to study optimal wavelength combinations for detecting Hb
3. to study correlations between multimodal neuroimaging measures
4. to study human blood-brain barrier disruption with multimodal measurements
48
49
4 Participants and methods
4.1 Participants
The studies were approved by the Regional Ethics Committee of the Northern
Ostrobothnia Hospital District. Written informed consent was obtained from all
participants prior to the fMRI scanning in accordance with the Helsinki declaration
and in addition to routine clinical BBBD information.
Study I
One 31-year-old male subject was used.
Study II
One 25-year-old male subject was used.
Study III
In this study, resting-state scans were performed in 11 healthy volunteers (3 women,
27.2±7.5 years old).
Study IV
Nine PCNSL patients (5 women, 55±16 years old) were monitored in 47
consecutive BBBD treatments.
4.2 Imaging, measurements and procedure
Study I
MRI was performed with Siemens Skyra 3T scanner using a 32-channel head coil.
Subject’s T1-weighted magnetization-prepared rapid acquisition with gradient
echo (MPRAGE) 0.9 mm voxel size anatomical magnetic resonance (MR) image
was used to reveal the subject’s exact head structure for accurate phantom
fabrication and experiment design. From the MR image’s sagittal slice a large vein,
50
sagittal sinus, was identified beneath the frontal skull and its thickness was
measured (4.0 mm) (Fig. 1). As can be seen, the distance between the top of the
skin and the surface of the vessel varied from 16 to 24 mm depending on the height
of the measurement point. The scalp, skull, CSF, grey matter and white matter were
also identified and their thicknesses measured (Fig. 2).
Fig. 1. Anatomical MR image. (Study I, published with permission from IEEE Photonics
Society.)
51
Fig. 2. Anatomical MR image of the brain for determining the thickness and shapes of
the various tissue layers in the forehead. (Study I, published with permission from IEEE
Photonics Society.)
Based on the subject’s anatomical MR image, two types of multi-layered phantoms
were fabricated at Optoelectronics and Measurement Techniques (OPEM) in the
University of Oulu: rectangular and head mimicking. Before the phantom
fabrication process, the reduced attenuation and scattering coefficients (Tuchin
2007) were calculated for each phantom layer from reference data presented in
(Sorvoja et al. 2010), using corresponding coefficients and g-factors. Both phantom
types consist of five layers mimicking the scalp, skull, CSF, grey matter and white
matter. Rectangular phantoms have a size of 4 cm x 4 cm and flat surfaces. Head
mimicking phantoms comprised layers with surface morphologies similar to the
shape of real brain layers taken from the axial anatomical MR image (Fig. 2). An
outline of the forehead (Fig. 3) was formed using that image, and a corresponding
metal mold of the phantom was fabricated.
52
Fig. 3. Multilayer mold for head-mimicking phantom. (Study I, published with
permission from IEEE Photonics Society.)
A tube was positioned inside the phantom mimicking a real blood vessel in the
brain in its size and localization. A large vein, identified earlier from the anatomical
MR image, was selected for our experiments. However, the main goal was only to
estimate light propagation depth, not to mimic exactly the blood flow in the vessel.
Furthermore, to simplify the placement of the vessel, it was decided to use two
constant depths: 19 mm, with the tube placed between the layers mimicking grey
and white matter; and 15 mm, with the tube placed between the grey matter and
CSF. The inner diameter of the tube was 4 mm and the thickness of the silicon layer
was 1.28 mm. Pulsations were generated using a WELCO WPX1 peristaltic pump,
and aqueous 20% intralipid suspension mimicked blood in the tube.
In order to monitor the cerebral cortex of the brain by NIRS, light should
penetrate (1-2 cm) into brain tissue through the scalp, skull and CSF to grey and
white matter and then be scattered back to skin surface. To estimate the penetration
depth by sensing pulsations in the tube inside phantoms, a measurement setup
presented in Fig. 4 was built. A 20% aqueous intralipid suspension was used for
pulsating flow and different fluid flow velocities were used. In the NIRS device,
two LEDs were used from the near-infrared region (wavelengths of 830 nm and
905 nm) with output powers of 2.4 mW and 0.4 mW in this order. The signals were
separated from each other by an amplitude modulation technique (Sorvoja et al. 2010).
53
Fig. 4. Phantom measurement setup. (Study I, published with permission from IEEE
Photonics Society.)
MC simulations were carried out to estimate light propagation in a multi-layered
medium with a complex geometry based on the methods presented in (Gorshkov &
Kirillin 2012). The main idea for MC simulations was to understand what the
measurement volume of the constructed NIRS setup is. It cannot be done in vivo,
but MC and phantom simulations allow us to build photon trajectory maps for a
predefined source-detector separation. These maps then demonstrate typical paths
of the photons contributing to the NIRS signal in the given configuration.
Rectangular phantom geometry and optical properties was used in our simulations
to fully correspond to the actual experiment. Table 1 presents the tissue layers of
the head used in the simulation and their optical properties. The absorption
properties of the layers at the corresponding wavelength (900 nm) were rather small
and were introduced by intrinsic absorption, without addition of absorbing ink. For
each simulation, 108 photons were employed.
54
Table 1. Thicknesses of the tissue layers and their optical properties at 900 nm
wavelength.
Layer L, mm µs`, 1/mm µa, 1/mm µs, 1/mm g n
Scalp 3 1.44 0.002 7.2 0.8 1.44
Skull 10 1.5 0.002 7.5 0.8 1.44
CSF 2 0.00001 0.0001 0.001 0.99 1.41
Grey matter 4 2.4 0.002 12 0.8 1.44
White matter 20 7.8 0.002 39 0.8 1.44
Study II
In this study, hardware- and software-based lock-in amplification of the NIRS
device was tested and compared by doing experimental measurements. Red (660
nm) and infra-red (830 nm) light of a high power light emitting diode (HPLED)
was used to illuminate the subject’s forehead. A 6 kHz sinusoidal carrier wave was
used in modulation to 830 nm and 8 kHz to 660 nm. Hardware-based lock-in
amplification was realized using an AD630 high-precision balanced modulator,
whereas software-based lock-in amplification was accomplished using LabVIEW.
Rather low carrier frequencies were used because of the data acquisition (DAQ)
card’s limited maximum scanning rate. For comparison, both methods used a
similar setup, which included a light source-detector pair, modulators, amplifiers,
filters and a DAQ card to save the captured data on a hard disk.
Different pairs of HPLEDs were also tested by using the software-based lock-
in amplification method. The subject’s forehead was illuminated at different
wavelength combinations using 3 cm source-detector distance, which has been
shown to be sufficient to enable sensing of pulsations from the brain cortex
(Myllylä et al. 2013b). Wavelengths of 660, 760, 810, 830, 850, 905, 940 nm were
used. Furthermore, the optical output spectrum of each HPLED was measured
separately using OceanOptics: Jaz EL 350 and USB4000-UV-VIS Miniature Fiber
Optic Spectrometer measurement devices. The obtained optical output spectra were
compared to the estimated optical absorption spectra of HbO and Hb, taken from
(Prahl 1999).
In the experiments, the breath-holding paradigm was used because it has been
proven to cause changes in the concentrations of HbO and Hb (Schelkanova &
Toronov 2011). BH generates blood CO2, which acts as a cerebral vasodilator and
causes increased CBF. CBF then increases both CBV and oxygenation. Our
measurement protocol lasted 4 minutes: the subject was first breathing normally
55
(60 s), then he was asked to hold his breath (30 s), breathe normally (60 s), hold his
breath (30 s) and breathe normally (60 s).
Study III
MRI was performed using Siemens 3T SKYRA with a 32-channel head coil and
utilizing the MREG sequence obtained from Freiburg University via close
international collaboration with Jürgen Hennig’s group (Lee et al. 2013, Zahneisen et al. 2012). It is a three-dimensional (3D) spiral, single-shot sequence that
undersamples 3D k-space trajectory, and hence enables faster imaging (Assländer et al. 2013). MREG samples the brain at 10-Hz frequency (TR = 100 msec, TE =
1.4 msec and flip angle = 25º) and offers thus about 20-25 times faster scanning
than conventional fMRI. Anatomical 3D MPRAGE (TR = 1900 msec, TE = 2.49
msec, flip angle = 9º, FOV = 240 and slice thickness = 0.9) images were used to
register the MREG data into 4 mm MNI space. Fingertip SpO2 and respiratory belt
data were also saved from the scanner.
An MRI-compatible NIRS measurement device has been developed since 2008
in cooperation between the Health and Wellness Measurement (HWM) group of
OPEM and Oulu Functional Neuroimaging (OFNI) group, both at the University
of Oulu (Sorvoja et al. 2010). The NIRS device allows the continuous measurement
of CBF of the brain cortex, especially quantifying concentrations of Hb and HbO,
in synchrony with fMRI. In this study, wavelengths of 660, 830 and 905 nm were
used. These wavelengths were chosen because both 660 nm and 830 nm are located
on both sides of the isosbestic point of the absorption spectrum of HbO and Hb.
That point is located ~800 nm, and there the extinction coefficient of both of these
is the same. The third wavelength was chosen to be 905 nm in order to probe
cytochrome aa3 activity, but it was not analyzed further in this study or in this
doctoral dissertation. One NIRS measuring channel was used and it was located on
the subject’s upper forehead to measure the DMNvmpf. The sampling rate was 10
kHz.
EEG and ECG were recorded using a 32-channel, MRI-compatible BrainAmp
system (Brain Products, Gilching, Germany) with Ag/AgCl electrodes placed
according to the international 10-20 system. The sampling rate was 5 kHz and
electrode impedances were <5 kΩ (Tallgren et al. 2005). Signals were band-pass
filtered from DC to 250 Hz. Baseline was recorded for 30 seconds with eyes open
and eyes closed outside the scanner room, testing signal quality before the actual
56
measurements. DC-potential measurements were made possible by removing skin
potential with stick abrasion technique (Vanhatalo et al. 2003).
Our MRI-compatible NIBP device was designed, implemented and developed
by the HWM and OFNI groups. It was used to measure vasomotor waves of arterial
BP. Cardiovascular pulses induce small movements of the skin, especially near the
arteries; these are sensed from the skin surface continuously and simultaneously
using two accelerometer sensors. One sensor is placed over the aortic valve on the
sternum and the other on the carotid artery on the neck, and PTT is calculated
between the sensors. It gives the transition time between the starting time of each
pressure pulse (opening of the aortic valve) and the ensuing diameter change of the
carotid. Further, PWV can be calculated from PTT once the distance between the
sensors has been measured. PWV is the speed of a pressure pulse propagating along
the arterial wall and since it depends on the pressure in the aorta (the higher the
pressure, the greater the velocity), systolic pressure in the aorta can be determined
by it. Moreover, BP changes caused by vasomotor waves are reflected also as
changes in PWV. The measurement method and its background are described in
more detail in the following reference (Myllylä et al. 2011). The sampling rate of
the raw acceleration signals was 10 kHz.
Anesthesia monitor signals (SpO2, CO2, ECG, and cuff-based NIBP) were also
registered using a 3T MRI-compatible anesthesia monitor (GE Datex-OhmedaTM;
Aestiva/5 MRI). The cuff-based NIPB was used for calibrating the continuous
NIBP before and after the actual measurements. Continuous measurement of SpO2,
CO2, and ECG data was optically transferred to a monitoring computer during
actual scans via a special server for time series with a sampling rate of 300 Hz. The
data were collected and saved in a monitoring computer by the Datex-Ohmeda S/5
Collect software. Both in this study and this doctoral dissertation, these data were
used only for verification purposes.
To enable accurate timing between all the different signals mentioned above,
they were synchronized with each other using the 3T SKYRA MR-scanner optical
timing pulse. It mediates millisecond-level trigger for synchronization of data
gathering for EEG, NIBP and NIRS. In addition, BrainAmp SyncBox was used to
verify that EEG amplifier and MR scanner are in synchrony via a transistor-
transistor logic pulse. Anesthesia monitor signals do not have millisecond-level
timings, but these are facilitated from the scanner artifact reflected in the ECG leads
and it works sufficiently well.
57
Study IV
The BBBD treatment for PCNSL in Oulu University Hospital is based on the
original procedure of Neuwelt and co-workers (Angelov et al. 2009, Neuwelt et al. 1979a), further developed here in Oulu in collaboration with Neuwelt’s group. On
the 1st treatment day, the patient is tested with routine clinical laboratory tests and
imaged with MRI or CT. In addition, Rituximab chemotherapeutic agent is given
for metabolic activation of the drug in the liver prior to the BBBD treatment. On
days two and three, the actual BBBD treatment is carried out with BBBD-enhanced
chemotherapy under general anesthesia.
Before anesthesia induction, intravenous (i.v.) phenobarbital and midazolam
are given and during the BBBD procedure, anesthesia is induced and maintained
using propofol. Two to three minutes prior to intra-arterial (i.a.) mannitol infusion
anesthesia is deepened up to EEG suppression level (entropy 0) by i.v. thiopental
bolus together with a benzodiazepine. At the same time, atropine is given to
counteract strong vasovagal effects of the BBBD. Muscle relaxants are not used
because they could encumber detection of clinical seizures caused by the infusions.
I.v. chemotherapy drugs are given 5-10 minutes prior to mannitol. The actual
BBBD treatment is administered to either one of the internal carotid arteries or to
the dominant vertebral artery. A hyperosmolar mannitol bolus is administered intra-
arterially after angiographic verification of the selected artery. Right after mannitol,
10-minute i.a. infusions of methotrexate (first) and carboplatin are given. I.v.
contrast medium is injected during i.a. chemotherapy infusions, and the degree of
BBBD is assessed using cone beam CT at the end of the procedure (Neuwelt et al. 1979b).
In this study, the same NIRS device was used as in earlier studies I-III. Each
subject was measured using one channel placed on the forehead beneath the EEG-
cap lead adjacent to Fp1 or Fp2 leads on the side of the infused artery. Wavelengths
of 660 nm and 830 nm were used, source-detector distance was 3 cm, and the
sampling rate was 1 kHz. DC-EEG and ECG were recorded using the same
BrainAmp system and Ag/AgCl electrodes as in study III.
58
4.3 Data pre-processing & analysis
Study I
In this study, raw NIRS time courses and their fast Fourier transform (FFT) spectra
were analyzed. FFT analyses were done using OriginPro software.
Study II
The measured raw NIRS time courses (for example, 660 nm and 830 nm data) were
converted into temporal changes of Hb and HbO concentrations using three
different Matlab programs, all based on MBLL. The value 5.93 was used as a DPF
factor for adult head (Van der Zee et al. 1992). Method 1 was based on equations
and specific extinction coefficients taken from Cope’s Ph.D. thesis (Cope 1991).
Method 2 was mostly based on an article by Boas (Boas et al. 2001). In that method,
the light absorption is calculated only for time points indicated by blood pulsation,
which means that the light intensity is compared only between the high and low
point of the received pulse (Zourabian et al. 2000). The third method was a freely
available software program, Homer (Huppert et al. 2009). The theory of MBLL has
been widely introduced in numerous publications (Boas et al. 2001, Cope & Delpy
1988, Cope 1991).
Study III
FMRI MREG data were processed with a typical FSL pipeline. 180 time points (18
seconds) were removed from the beginning to minimize T1-relaxation effects.
Head motion was corrected using FSL MCFLIRT software (Jenkinson et al. 2002).
FSL’s brain extraction tool (BET) was used for optimization of the deforming
smooth surface model (Smith 2002). Spatial smoothing was done with fslmaths 5
mm full width at half maximum (FWHM) Gaussian kernel. 3D MPRAGE images
were used to register the MREG data into MNI space at 4 mm resolution prior to
group ICA as a standard procedure in MELODIC.
To separate RSNs from noise sources, group probabilistic ICA (PICA) was
calculated for MREG data as a part of normal MELODIC procedure (Beckmann &
Smith 2004). Model order was chosen to be 70 as an initial step, and previous
criteria (Kiviniemi et al. 2003, Kiviniemi et al. 2009) were used to separate RSNs
from noise components. 47 independent components (ICs), which were considered
59
to be RSNs, and 23 motion- and CSF-pulsation-related noise components were
detected. The dominant DMNvmpf component beneath the frontal NIRS sensor was
selected for further analyses.
Dual regression was used to generate subject-specific versions of the spatial
maps and associated time series from the spatial maps of the group-average analysis
(Beckmann et al. 2009, Filippini et al. 2009). The first stage of dual regression
regresses the group-spatial-maps into each subject’s 4 dimensional (4D) dataset to
give a set of time courses. The set of these subject-specific dual-regressed time
series, one per group-level DMNvmpf spatial map, were used in the following
correlation analyses. The dual-regression-derived time signals were used since the
time domain signal can be analyzed without any further need of corrections for
auto-correlation effects (Boyacioglu et al. 2013).
NIRS data analyses were done in similarly as in study II (method 1) using
Matlab. Wavelengths of 660 nm and 830 nm were used.
EEG recordings were offline post-processed using the same procedure as in
Hiltunen et al. (Hiltunen et al. 2014). The ballistocardiographic (BCG) and gradient
artifacts were corrected using the average artifact subtraction method (Allen et al. 1998, Allen et al. 2000). ICA was calculated to EEG data in Matlab with EEGLAB
(Delorme & Makeig 2004) using the Infomax algorithm (Bell & Sejnowski 1995)
with sub-Gaussian source detection. The maximum number (32) of ICs was used
to get maximally temporally comparable ICs to MREG RSNs.
In NIBP data analyses, the highest peaks of both accelerometer sensor data
were identified by first searching for the maximum value of each time interval
between heartbeats. Next, delays between the corresponding peaks were calculated.
The method for finding heart peaks and validating the acceleration response
followed by determination of the PTT and PWV values was performed using
Matlab and it is presented in more detail in the following references (Myllylä 2014,
Myllylä et al. 2012, Myllylä et al. 2011).
For anesthesia monitor data (ECG, CO2 and SpO2), FFT was done to obtain the
power spectra of the signals. These power spectra were used to determine each
individual’s respiration and cardiac frequency bands to avoid those in our VLF
analysis. Cuff-based BP signal was used to calibrate our NIBP device.
All of the data were first down-sampled to correspond with the sampling rate
of MREG data (10 Hz) and time signals were detrended by fitting a line and
removing it. Zero-lag correlations between all different modality full band (FB)
signals were calculated with Pearson correlation coefficients and absolute values
were averaged between subjects. All possible combinations were compared.
60
Study IV
Temporal changes of Hb and HbO concentrations were calculated from raw NIRS
time courses using Matlab’s NIRS processing package called HomER2 (Huppert et al. 2009). Hb and HbO time courses were then low-pass filtered with the cut-off
frequency at 0.15 Hz and down-sampled to 10 Hz.
EEG data were referenced to the common average and linear drifts were
removed from all channels. Thereafter, signals were down-sampled to 10 Hz (anti-
aliasing with a FIR filter) and low-pass filtered using a 21-point moving average
filter for detection of infraslow EEG signals (i.e., DC shifts). The specimen trace
in (Fig. 16) is shown using a wider bandwidth (low-pass filter cut-off at 48 Hz).
Average responses shown in (Fig. 17) were calculated for left anterior, right anterior,
left posterior and right posterior channels. When re-referencing to the ECG
reference was done, the ECG reference electrode signal was low-pass filtered like
the DC-EEG channels. EEGLAB (Delorme et al. 2011) was used for topographic
illustrations of DC-EEG data based on all the 31 recorded channels of the 10-10
system.
61
5 Results
5.1 NIRS measurements
5.1.1 Light propagation (Study I)
Light propagation was studied by changing the source-detector distance in the
NIRS device. This was applied by means of phantom measurements, in vivo measurements of the subject and MC simulations.
Phantom measurements
The effect of changing the distance between the source and the detector was studied
by measuring the pulsating liquid in the rectangular phantom as described in Fig.
4. Particularly, the interest was to find out whether it is at all possible to optically
sense movements at depths of 15 and 19 mm. In the measurement, the tube was
always in the middle between the source and the detector and the direction of the
source-detector pair axis was perpendicular to the tube.
The PPG reference measured from the surface of the silicone tube at a source-
detector distance of 2 mm always gave an accurate signal response and was used
to observe whether NIRS was able to sense the same response simultaneously (Figs.
5-6). Fig. 5 shows NIRS responses when using source detector distances of 3 cm
(A) and 4 cm (B) and depth of 19 mm. In this case, 3 cm was sufficient to sense
movement, but by increasing the source-detector distance, the response became
clearly better. For the depth of 15 mm, 2.5 cm was the minimum source-detector
distance to sort of sense the pulsation (Fig. 5A.). However, increasing the distance
from 2.5 cm (A) to 3 cm (B) and 3.5 cm (C) clearly improved the response,
especially in the case of 905 nm (Fig. 6.). Pulsations were not constant so the delay
between the peaks of PPG and NIRS signals varied, but in any case, corresponding
peaks could be easily recognized visually from both figures.
62
Fig. 5. Phantom measurements of a pulsation liquid under the grey matter at depth of
19 mm using wavelengths of 830 nm and 905 nm. Source-detector distance is (A) 3 cm
and (B) 4 cm. (Study I, the figure is combined from two figures published in Study I,
published with permission from IEEE Photonics Society.)
63
Fig. 6. Phantom measurements of a pulsation liquid above the grey matter at depth of
15 mm using wavelengths of 830 nm and 905 nm. Source-detector distance is (A) 2.5
cm, (B) 3 cm and (C) 3.5 cm. (Study I, the figure is combined from three figures published
in Study I, published with permission from IEEE Photonics Society.)
64
In vivo measurements of the subject
These measurements were performed with the same person whose forehead brain
tissue layer thicknesses were retrieved for phantoms and the MC simulations. In
addition, the same measurement setup and NIRS device were used as in the
phantom measurements to enable comparison. Fig. 7 shows spectrum signal
responses using wavelengths of 830 (A) nm and 905 nm (B) at three different
source-detector distances (2 cm, 2.5 cm, and 3 cm).
Fig. 7. In vivo measurements from the subject’s forehead at wavelengths of (A) 830 nm
and (B) 905 nm using source-detector distance of 2 cm, 2.5 cm and 3 cm. (Study I, the
figure is combined from two figures published in Study I, published with permission
from IEEE Photonics Society.)
65
Of interest in Fig. 7 are especially LFOs of cerebral hemodynamics and
metabolism in the frequency range between 0.01-0.15 Hz. In this spectral range,
peaks are often visible when measuring signal from the brain by fMRI or NIRS.
Comparison of frequency responses between the three source-detector distances
shows that at 3-cm separation distance there are peaks at frequency band of 0.01-
0.15 Hz but not at shorter distances. This indicates that by increasing the source-
detector distance to 3 cm, illuminated light starts to reach the cerebral cortex at both
wavelengths. This is in correlation with the phantom measurements.
Monte Carlo and phantom simulations
Fig. 8 presents comparison of experimental and simulated curves for attenuation of
NIRS signal at wavelengths of 830 nm and 905 nm as a function of source-detector
distance. For the rectangular phantom, one can see quite good agreement between
simulation and experiment. The agreement between the simulations and the
experiment results demonstrates the adequacy of the chosen parameters.
The photon trajectory maps obtained from MC simulations are shown in Fig.
9. From this figure, one can see that it is possible to measure signal from the brain
(grey matter and even white matter) using a source-detector distance of 30 mm (A,
C) and 40 mm (B, D). In addition, the measurement volume for source-detector
separation of 40 mm is located at a larger depth from the sample boundary
compared to that of 30 mm, which ensures deeper sensing and lower contribution
of upper layers to the NIRS signal. Both of these observations are in good
agreement with our experimental studies.
66
Fig. 8. Dependence of the NIRS signal at 830 nm and 905 nm on the source-detector
distance for the rectangular phantom: comparison of experimental results with MC
simulations. (Study I, published with permission from IEEE Photonics Society.)
67
Fig. 9. Photon trajectory maps (logarithmic color map) for 830 nm (A,B) and 905 nm
(C,D) for source-detector distances of 30 mm (A,C) and 40 mm (B,D). White lines indicate
the boundaries of the phantom layer. (Study I, published with permission from IEEE
Photonics Society.)
5.1.2 Different wavelength combinations/flexibility (Study II)
HPLEDs are suitable for recording blood oxygenation. Several wavelength
combinations can be used, but the combination of 830 nm and 660 nm gave the
best SNR and distinct separation of Hb in the human forehead. Fig. 10 shows
examples of Hb concentration changes using different wavelength combinations
and breath holding paradigm.
68
Fig. 10. Concentration changes of Hb measured with different wavelength
combinations. (Study II, published with permission from Elsevier.)
5.2 Multimodal measurements (Study III)
5.2.1 Technical realization and integration
The technical line-up of our multimodal “Hepta-scan system” is illustrated in Fig.
11. With Hepta-scan, seven key physiological variables (MREG, NIRS, EEG, NIBP,
SpO2, CO2 and ECG) known to affect brain hemodynamics can be simultaneously
measured with high sampling rate. The MR-scanner mediates a millisecond-level
optical timing pulse which acts as a trigger for NIRS, EEG and NIBP to enable
exact synchronization. Furthermore, the BrainAmp Syncbox was used to ensure
that the EEG amplifier and the scanner are in synchrony via a transistor-transistor
logic pulse (Hiltunen et al. 2014). Anesthesia monitor signal timing is facilitated
from the scanner artifact reflected in the ECG-leads, enabling approximately half a
second timing accuracy for all other devices. Monitoring data of all patients are
optically transferred from the MR-room via a wave tube to laptops in the control
69
room. The measured example raw signals from one subject are presented in Fig.
12B.
Fig. 11. Demonstration of the measurement setup. The placement of electrodes and the
anesthesia monitor used are presented on the left inside a shielded room. All signals
are transferred outside optically via a wave tube. Color codes for the signals: EEG, blue;
NIRS, red; NIBP, green; anesthesia monitor data, violet; timing pulse, black dashed line.
Additionally, respiration from belt and fingertip SpO2 were measured by the scanner
itself. Scanner timing pulse triggered EEG, NIRS, and NIBP. EEG was also synchronized
with the scanner’s clock signal via SyncBox device. On the right, the raw signals from
each measurement device are shown. The starting point of the measurement can be
clearly seen from the beginning of the scanning artifact in the EEG signal. (Study III,
published with permission from Mary Ann Liebert, Inc. publishers.)
70
Fig. 12. Example data from a single subject’s (A) z-statistic map of the selected MREG
DMNvmpf group PICA component registered to the subject’s anatomical brain image
(sagittal, axial, and coronal slices). (B) Ten-minute FB time series of DMNvmpf (black),
selected EEG IC (blue), NIRS-Hb (red), and NIBP (green). (C) VLF time series of signals
presented in (B). (D) and (E) are the enlarged sections (2 min) of the original signals
indicated by the black block in (B) and (C). The table presents correlation coefficient
values between different modality signals presented in (B-E). (Study III, published with
permission from Mary Ann Liebert, Inc. publishers.)
5.2.2 NIRS, EEG and NIBP correlations to MREG DMNvmpf
Correlations were calculated between the selected MREG IC component (DMNvmpf)
time course and other modalities (NIRS, EEG IC, and NIBP). Z-statistic map of the
selected MREG DMNvmpf component is presented in Fig. 12A. Two different time
window lengths (10 min and 2 min) and frequency bands (FB and VLF) were used
in correlation calculations. Examples of all of these signals are presented in Fig.
12B-E. The table in Fig. 12 shows one subject’s individual correlation coefficients
between these signals.
Average correlations and corresponding standard deviations (SDs) between
DMNvmpf and selected EEG IC were 0.25 (0.06), NIRS-Hb 0.14 (0.14), NIRS-HbO
71
0.12 (0.11) and NIBP 0.11 (0.11) in 10-min measurement using full-frequency band
(Fig. 13). The respective correlations and SDs in the VLF band were 0.30 (0.06),
0.24 (0.17), 0.18 (0.13) and 0.11 (0.08). All aforementioned 10-min correlations
were significantly (p<0.05) lower than the highest 2-min ones.
Fig. 13. Average correlations and their SD bars between MREG DMNvmpf and EEG IC
(blue), DMNvmpf and NIRS-Hb (red), DMNvmpf and NIRS-HbO (orange), and DMNvmpf and
NIBP (green) both on whole 10-min measurement (left) and the highest 2-min time
window (right) and both on FB and on a VLF band. VLF bars are lighter in color. All VLF
correlations that are significantly higher than the corresponding FB correlations are
marked with a star. All 2-min correlations are significantly higher than the
corresponding 10-min ones. (Study III, published with permission from Mary Ann
Liebert, Inc. publishers.)
Fig. 13 presents absolute values for average correlations between different
modalities and DMNvmpf signal. To further demonstrate the temporal variability of
the intermodality correlations, the highest positive and lowest negative 2-min time
window average correlations are also presented in Fig. 14. The table in Fig. 14
shows the mean correlations: the highest positive correlations range from 0.35 to
72
0.09 while the lowest negative correlations range from -0.18 to -0.09 on FB. On
VLF, the highest positive correlations varied between 0.54 and 0.35 and the lowest
negative correlations between 0.04 and 0.41. Only significant (p < 0.05) differences
between highest positive average correlation coefficients and absolute values of
lowest negative correlation coefficients were found between DMNvmpf and EEG IC
correlations (both FB and VLF). It needs to be noted that there was no significant
correlation between the order number of the 2-min time window versus the number
of maximal correlation hits in either VLF (F = 3.09836, p = 0.12176) or FB data (F
= 1.3527, p = 0.28292).
Fig. 14. Average correlations (both highest positive and lowest negative) and their SD
bars between MREG DMNvmpf and EEG IC (blue), DMNvmpf and NIRS-Hb (red), DMNvmpf
and NIRS-HbO (orange), and DMNvmpf and NIBP (green) on FB (left) and on VLF (right).
The table presents average correlation coefficient values (highest positive [ + ] and
lowest negative [ - ]) and their SD. Significant difference between the highest positive
and absolute value of the lowest negative is marked with a star. (Study III, published
with permission from Mary Ann Liebert, Inc. publishers.)
5.2.3 EEG/NIRS/NIBP signal correlations with each other
Average correlations were also calculated between EEG ICs, NIRS, and NIBP
signals with two different time window lengths (10 min and 2 min) and each with
two different frequency bands (FB and VLF). Average correlation coefficients and
their SDs are illustrated in Fig. 15. All calculated correlations were higher in VLF
than in FB, and also in 2-min time window than in whole 10-min measurement.
73
Furthermore, all differences were statistically significant except for two cases (10-
min VLF vs. FB in correlations between EEG-NIBP and Hb-NIBP). The highest
correlations were always seen between EEG IC and Hb. Notably, the correlation
between EEG and Hb exceeds those between EEG and MREG in the 2-min time
window.
Fig. 15. FB and VLF average correlations and their standard deviation bars between
EEG IC and NIRS-Hb (black), EEG IC and NIRS-HbO (blue), EEG IC and NIBP (green),
NIRS-Hb and NIBP (red), and NIRS-HbO and NIBP (orange) both on whole 10-min
measurement (left) and highest 2-min time window (right). VLF bars are lighter in color.
All VLF correlations that are significantly higher than the corresponding FB correlations
are marked with a star. All 2-min correlations are significantly higher than the
corresponding 10-min ones. (Study III, published with permission from Mary Ann
Liebert, Inc. publishers.)
74
5.3 Real-time monitoring of human BBBD (Study IV)
5.3.1 NIRS
Real-time monitoring of human BBBD by NIRS revealed that both HbO and Hb
signals plummeted when the 30-s mannitol infusion started (Figs. 16 and 17). The
drop was caused by the dilution of blood and it was quickly followed by a
hyperemic increase in cortical oxygenation (30 s-2 min) and an increase in HbO to
a substantially elevated level. The increase in HbO persisted until the end of the
20-minute measurement period, showing a slow and rather linear trend towards the
original baseline level. The Hb signal behaved differently; after the initial fall
induced by the mannitol bolus, Hb increased temporarily towards its original level,
but soon started to fall again, obtaining a steady level clearly below the original
baseline. This prolonged decrease below the original baseline lasted for more than
20 minutes, showing a very slow recovery towards the original pre-bolus level.
The NIRS measurements were always performed on the forehead due to
imaging- and procedure-related limitations during the BBBD. The results were
concordant between both carotid artery BBB disruptions, and the very prolonged
cerebrovascular response was strikingly different compared to the gradually fading
DC-EEG shifts. Obviously, a frontal NIRS cannot monitor the vertebral artery
territory, and therefore the much less prominent and delayed NIRS responses
evoked by the vertebral artery mannitol infusion are shown in (Fig. 17).
5.3.2 EEG
The mannitol infusion induced a huge multi-phasic potential response in the EEG
channels monitoring the affected arterial territory, with amplitudes that were orders
of magnitude larger (up to 2 mV) than those of commonly observed EEG rhythms.
For illustration of typical responses seen in the middle of the infused right carotid
arterial territory, a representative trace recorded at F4 (or F3 for the left carotid
artery infusion) during the BBBD procedure with right carotid artery infusion is
shown in (Fig. 16). The response to the 30-s mannitol infusion commenced with a
negative peak lasting 1.5-2 minutes. The negative peak was followed by a slow
potential descent below the pre-bolus level. Grand average DC-EEG traces at
opposite quarters of the scalp were calculated to illustrate characteristic responses
evoked by i.a. mannitol infusions (Fig. 17). In all cases, the responses were fading
at the end of the 20-min monitoring of the mannitol effect.
75
Fig. 16. Characteristic raw EEG (upper graph) and NIRS (lower graph) responses seen
simultaneously during the BBBD procedure of right carotid artery mannitol infusion.
The mannitol infusion induces a robust negative shift in the EEG which is followed by
a slow potential descent below the pre-bolus level. Boxes (a-d) illustrate zoomed parts
of an EEG signal. NIRS signals (HbO = red solid line and Hb = blue dashed line) plummet
due to dilution of blood when the 30-s i.a. mannitol infusion starts. Subsequently, both
of them start to increase towards the original levels. HbO rises above the baseline and
stays there for more than 15 minutes whereas Hb approaches the baseline level but
soon starts to fall again, obtaining a steady level clearly below the original baseline.
76
Fig. 17. Grand average DC-EEG (upper graphs) and average NIRS (lower graphs) traces
(shaded area indicates values within ±1 SD) illustrating characteristic responses
evoked by i.a. mannitol infusion. Mannitol infusion starts at time = 0 and lasts 30 s. Left
(n = 13) and right (n = 16) carotid artery infusions induce a negative DC-EEG shift in
electrodes above the treated arterial territory, which outlasts the infusion and is
followed by a slower shift of opposite polarity. Contralateral posterior electrodes record
a response that is qualitatively similar but reversed in polarity. A clear fall in the NIRS
signals is seen during left (n=8) and right (n=8) carotid artery infusions of mannitol,
followed by a pronounced rise in HbO and a transient partial recovery of Hb, after which
Hb settles down to a level below the original baseline and HbO decreases slowly but
does not fully recover by the end of the 20-min monitoring period. Vertebral artery
infusions show a fronto-occipital DC-EEG potential shift (n = 18) without a lateralized
effect, as expected. Re-referencing to the distant ECG reference electrode reveals that
there is a negative shift throughout the entire scalp. The early transient shifts in the
NIRS signals shown for vertebral artery infusions (n = 7) are more delayed and have
much smaller amplitudes because NIRS was always measured on the forehead, i.e.,
they show responses generated in a non-infused brain area. After 5 minutes, the Hb and
HbO traces cross, HbO remains below the original baseline, whereas Hb shows a near-
complete recovery.
77
6 Discussion
The research of this thesis is based on verification of NIRS’ capability to measure
targeted brain physiological phenomena, extending to control and patient
monitoring. The minimum source-detector distance and optimal wavelength
combination were defined. Thereafter, the optimized NIRS system was integrated
as a part of our ultra-fast, multimodal neuroimaging setup, Hepta-scan, in
millisecond synchrony. The synchronous critical sampling revealed significant
dynamics in correlations between multimodal brain scan data already at individual
level. Consequently, the novel multimodal physiological measurements showed
pioneering findings, such as glymphatic brain pulsations and the capability to
monitor blood-brain barrier opening during therapeutic BBBD with NIRS and DC-
EEG in humans.
6.1 Properties of the NIRS device
Our extensive results show that it is possible to measure signal from the brain’s
cerebral cortex using different source-detector distances and different wavelength
combinations in the NIRS device. The minimum source-detector distance in adults
should be at least 2.5 cm, which is in good agreement with the literature (Gibson et al. 2005, Rossi et al. 2012). However, increasing the distance to 3 cm gives a better
signal. Actually, 3 cm is the most often used source-detector distance in commercial
devices (Ferrari & Quaresima 2012, Scholkmann et al. 2014). Moreover, based on
phantom measurements, it seems that 4 cm gives an even better signal to infrared
wavelengths, especially to 905 nm. The reason for this may be that at these
wavelengths, thanks to increased depth in light propagation, SNR is more
optimized for this particular placement depth of the tube.
Our results also show that the combination of 660 nm and 830 nm gives the
most accurate results, and therefore it has mainly been used in our studies. This
finding is also in good agreement with the literature. The wavelength of 690 nm
was not used even though it has earlier been shown to be good (Kawaguchi et al. 2008, Sato et al. 2004) because a suitable HPLED does not exist.
At present, commercial NIRS devices already have multiple channels whereas
the advantages of our NIRS device are that it is adaptive, fMRI-compatible and has
the availability to use different, selectable wavelengths, depending on the purpose
of the study. In addition, the number of wavelengths can be chosen. As Scholkmann et al. pointed out, when measuring only HbO and Hb a higher SNR is achieved by
78
using a few discrete, but well-chosen wavelengths (Scholkmann et al. 2014). In
commercial devices you cannot change either of these, and the selections usually
are something of a compromise.
6.1.1 Different analysis methods in Study II
Lock-in amplification enables a higher SNR ratio and coherent detection as a result
of background light suppression, noise reduction and wavelength-encoding. The
lock-in amplification used in our NIRS data analysis is based on the method
described, for instance, in the article by Scofield (Scofield 1994). Different
wavelengths are distinguished from each other before lock-in amplification by
using narrow band-pass filters. It is possible to adjust the phase and frequency of
the reference signal used for different wavelengths. We have used opposite phase
in the lock-in demodulation. Therefore, negative signal values are obtained from
the output of lock-in amplification. Next, the signal is low-pass filtered in order to
remove second harmonic. Because our result signal is negative it is multiplied by
the value minus one before calculating the optical density values.
In Study II we compared different methods for calculating Hb and HbO. In
method 2, the DC component of the detected signal was removed for detection of
the minimum signal value of each pulse, which in this case was the highest negative
value. The determined minimum and maximum voltage values were then squared
in order to change them to correspond to power values and to change the minimum
value from negative to positive. However, it was later noticed that this procedure,
in particular the squaring, may be incorrect. The influence of the squaring was later
evaluated, and fortunately its effect found to be minimal. The only difference was
that the absolute values of Hb and HbO were two times bigger than without the
squaring. However, both the shape and SNR of the result signal did not change.
Nevertheless, method 2 gave lower correlation than method 1 when compared to
the Homer method, and its use would thus require further studies. Because of this,
it was decided that method 1 and the Homer method would be taken into use in our
further studies.
6.2 Feasibility of multimodal imaging
We developed a laboratory for simultaneous multimodal measurements to study the
constituents of spontaneous RS brain activity. Even though RS brain activity has
become a useful approach to analyze brain function (Friston 2011), the origin of
79
BOLD signal LFOs is still relatively unknown. It has been suggested that it might
arise from neuronal activation through neurovascular coupling (Hiltunen et al. 2014, Huster et al. 2012, Laufs 2012), from global systemic fluctuations (Tong et al. 2015, Tong et al. 2013), from metabolic sources (Heekeren et al. 1999, Obrig et al. 2000, Vern et al. 1997), or from a combination of all three (Fox & Raichle 2007,
Kiviniemi 2008). Furthermore, changes in respiration and HRV have also been
shown to alter the BOLD signal (Birn et al. 2008, Chang et al. 2013, Wise et al. 2004).
In this thesis, our multimodal setup “Hepta scan” was implemented for
simultaneously measurable modalities, fMRI, NIRS, EEG, NIBP and anesthesia
monitor signals, all in millisecond synchrony. Importantly, the responses recorded
with the different methods are not influenced by each other but are totally
independent. Hepta scan enables measuring the human brain 20 times faster and in
a more versatile manner than before. Accordingly, it offers new perspectives to the
understanding of brain activity that cannot be understood by measuring any of the
multimodal data alone. To our knowledge, the Hepta scan together with ultrafast
MREG sequence is one of the few critically sampling multimodal neuroimaging
set-ups in the world.
The basic idea of simultaneous multimodal imaging was to find phase lags that
could indicate which signal source leads and which follow. But instead, a new form
of temporal dynamics in the correlations between the multimodal data was
discovered: remarkable non-stationarity between different modalities. Moreover,
simultaneous multimodal measurements showed that VLF MREG signal
fluctuations are more linked to DC-EEG signal source than other sources, but also
supported the hypothesis that LFOs reflect the simultaneous activity of several
physiological fluctuations. So at the present stage, with the critically sampled
multimodal data used, the origin of spontaneous brain activity fluctuations seems
to be a dynamic interaction of several physiological factors.
Using long measurement periods, inter-modal correlation coefficients are
relatively low (0.2-0.3) because of the observed non-stationarity. On the other hand,
in shorter time windows correlation coefficients can be very high (> 0.7). The ultra-
fast scanning data, with increased statistical power and no aliasing, strongly
suggests that the non-stationarity is most likely a real phenomenon. Based on this,
it seems that everything in the brain is fluctuating separately. There seem to be
intermittent moments of integrated coherence in almost every measured index. This
coherence may be caused by a factor still to be discovered.
80
In addition, with the help of the Hepta scan data developed in this thesis, the
cerebrospinal fluid pulsations in the human brain were studied with MREG and
three types of physiological mechanisms affecting them were detected: cardiac,
respiratory, and VLF pulsations (Kiviniemi et al. 2016). The cardiac pulsations
induce a negative MREG signal change in peri-arterial regions that extends
centrifugally and covers the brain in about 1-Hz cycles. The respiratory pulsations
(~0.3 Hz) are centripetal periodical pulses that occur dominantly in peri-venous
areas. The third types of pulsations are very low frequency (0.001–0.023 Hz) and
low frequency (0.023–0.73 Hz) waves, both of which propagate with unique
spatiotemporal patterns. Our findings open new views into cerebral fluid dynamics.
The glymphatic brain pulsations may explain some of the dynamic variability in
the inter-modality correlations. In the future, our aim is to extend knowledge about
glymphatic pulsations based on multimodal imaging and especially NIRS.
Moreover, preliminary results on glymphatic pulsations indicate that in the future,
they might be very useful even in clinical analysis.
6.3 Correlation analysis and non-stationarity of multimodal signals
Remarkable temporal variability was found from correlation values between both
MREG and other modalities as well as between physiological signal correlations
without MREG (Figs. 13, 15). Furthermore, all modalities presented both positive
and negative correlations to MREG in a 2-minute time window (Fig. 14) and
consequently, all of the multimodal signals had higher correlation coefficients in
that window compared to full 10-minute measurement (both FB and VLF).
Interestingly, non-stationarity was found not only from FB but also from VLF band,
even though all of our modalities are critically sampled. Due to critical sampling,
all the high frequencies were precisely filtered out in the VLF band without aliasing.
This implies that none of the physiological signals are stable over time, so all affect
the MREG signal differently at different time points. The reason for this new
discovery is not yet known.
To the best of our knowledge, this study was the first where ultrahigh-temporal-
resolution fMRI and NIRS signals were studied together, and these results are
therefore unique. Interestingly, both Hb and HbO correlated almost equally to
MREG signal on average. Hb correlated generally a little bit better while HbO
correlated better in a 2-minute time window for VLF signals. This is well in line
with the literature as earlier fMRI-NIRS studies have reported both types of
correlations. In general, conventional BOLD has been reported to correlate better
81
with Hb (Huppert et al. 2006, Toronov et al. 2003), but also opposite opinions has
been presented (Duan et al. 2012, Strangman et al. 2002b).
Our results showed also that the MREG signal had the highest correlation
together with EEG (Fig. 13). Moreover, there were significant differences between
the highest positive and lowest negative correlation values between MREG and
EEG in a 2-minute time window whereas none were seen between MREG and both
NIRS and NIBP t(Fig. 14). This may be due to the selection procedure of the EEG
IC signal because the EEG IC component was always chosen so that it gave the
best positive correlation to the selected MREG IC signal in a 10-minute time
window. However, this suggests also that the origin of the MREG signal is mainly
electrophysiological.
The emergence of non-stationarity in our data may be related to increased data
sampling rate of fMRI in the case of FB as the data now have richer physiological
noise reducing the correlation. On the other hand, the VLF data are much better
filtered due to critical sampling and do not have aliased physiological signals such
as respiration and cardiac cycle. All in all, non-stationarity is a new finding. The
reason for this variability is unknown and calls for further research, e.g. with
correlation phase analytics.
However, one reason for the observed non-stationarity could be cerebral
cerebrospinal fluid pulsations known as glymphatic pulsations (Kiviniemi et al. 2016). Earlier, these quasi-periodic activity patterns were linked to local field
potential spread over the brain cortices in time both in rats and in human subjects
in VLFs (Kiviniemi et al. 2016, Majeed et al. 2011a, Thompson et al. 2015).
Another reason could be local adaptive vasomotor control phenomena as suggested
by Sirotin and Das (Sirotin & Das 2009). The localized vasomotor changes
dominate on task/stimulus-state transitions, which may well also exist in resting
DMN. Metabolic oscillations might also have an effect because they have been
shown to be functionally connected, at least in some studies (Obrig et al. 2000,
Vern et al. 1997).
6.4 Real-time monitoring of human BBBD
Our clinical results demonstrate continuous, real-time monitoring of human BBBD
using NIRS and scalp-recorded DC-EEG for the first time. BBBD was induced
with i.a. mannitol infusion in anesthetized patients receiving chemotherapy for
PCNSL. Since the intact BBB makes the brain parenchyma inaccessible to
82
hydrophilic drugs, new diagnostic and therapeutic innovations involving BBBD
may benefit from the methods described here.
The BBB forms very effective protection of the brain against pathogens present
in blood. On the other hand, it also inhibits the penetrance of hydrophilic drugs into
brain tissue, which is problematic in many treatments. Therefore, an increasing
amount of interest has been paid to investigating the role of BBB in brain pathology
and to treating brain diseases over the past few years (Neuwelt et al. 2011,
Obermeier et al. 2013). PCNSL is one of the most aggressive brain tumors,
typically treated using methotrexate (Angelov et al. 2009, Kasenda et al. 2015).
Even with high-dose i.v. methotrexate or intra-thecal chemotherapy, mortality is
high and progression-free survival time is counted in months (Angelov et al. 2009,
Kasenda et al. 2015). Tumor growth affects BBB integrity locally, enabling
systemic chemotherapy. However, the lymphoma cells form micro-metastases into
areas behind intact BBB, which inhibits the penetrance of hydrophilic medications
into the otherwise chemo-sensitive cells, and thus hinders effective use of
treatments. Therefore BBBD has been combined with methotrexate chemotherapy
in order to treat PCNSL (Angelov et al. 2009, Doolittle et al. 2014, Neuwelt et al. 1979a). To this end, i.a. hyperosmolar mannitol infusions are safely used to
transiently disrupt the BBB, which increases the brain penetration of
chemotherapeutics by up to 100-fold and thereby improves the response to
treatment (Angelov et al. 2009, Doolittle et al. 2008, Neuwelt et al. 2011, Neuwelt et al. 1979a).
The ability to monitor the degree and duration of BBBD is crucial for the
treatment of PCNSL because it has a direct link to patient outcome (Doolittle et al. 2001). Both sub-optimal and excessive BBB opening increases mortality and
complications (Kraemer et al. 2001). Evidently, means for real-time monitoring of
the degree of BBBD would be highly beneficial but previously no quantitative,
real-time methods existed. Our present data show that the readily applicable
method of NIRS and DC-EEG offers the possibility to monitor the level of BBBD
in real time.
Real-time measured NIRS data showed a remarkable triphasic cerebrovascular
response during carotid artery infusions (Fig. 16). It began with a rapid fall in Hb
and HbO, consistent with dilution of blood during the 30-sec mannitol infusion.
This was followed by a hyperemic second phase (30 sec–2 min) showing markedly
elevated HbO and a momentary imperfect recovery of Hb. Subsequently, within a
minute, Hb fell again and HbO showed a further increase, resulting in only partial
recovery whereas the DC-EEG signals largely recovered by the end of the 20-min
83
monitoring period. Notably, the Hb and HbO changes after the early bolus effects
were almost linear, in contrast to the non-linear DC-EEG changes. The prolonged
fall in the Hb level cannot be explained merely by hyperemia-induced dilution;
rather, it reflects temporary cessation of oxygen consumption since Hb is not being
produced. This might well be a sign of BBB opening.
Moreover, simultaneously measured DC-EEG showed a huge mV-level
negative shift in channels monitoring the affected arterial territory during carotid
artery infusions, which also suggests BBB leakage. The tight BBB maintains a
trans-endothelial voltage between blood and brain tissue (HELD et al. 1964, Revest et al. 1993, Woody et al. 1970). Normally, the human brain is positive with respect
to blood by 1 to 5 mV (Sørensen et al. 1978), and changes in this potential can
cause up to mV-level shifts in human scalp (Vanhatalo et al. 2003, Voipio et al. 2003).
In summary, both NIRS and DC-EEG findings support the conclusion that the
changes do indeed reflect BBB disruption and its subsequent gradual sealing, and
not responses of the intact BBB to changes in brain hemodynamics. Importantly,
the NIRS and DC-EEG data can be used as a biomarker of BBB status in disruption
therapy. In the future, more advanced analysis may reveal novel diagnostic
applications for BBB integrity with such multimodal wearable devices.
6.5 Future directions and technical improvements
A special multimodal laboratory in an MRI environment was implemented during
this doctoral thesis. It has already been used for over 150 subjects including
different patient groups, such as epilepsy, dementia and trauma patients. In the
future, we will measure and analyze more patients with different brain disorders.
Furthermore, we will endeavor to develop new biomarkers for brain disorders such
as epilepsy and narcolepsy.
Our future aim is also to develop further the utilized NIRS method. Currently,
a new MRI-compatible NIRS device with four measuring channels and selectable
wavelengths between 600-1500 nm is used. With the new device we can use the
receiver at a short distance; this will have significant influence on reducing
extracerebral contribution to the signals (Virtanen et al. 2009). Moreover, we can
compare side differences of NIRS signals between infused and non-infused artery
in the case of future BBBD measurements. However, we still want more channels
and want to integrate NIRS with EEG, as some study groups have already
successfully done (Lareau et al. 2011, Sawan et al. 2013). This development will
84
further enhance the possibility to study multimodal signal interactions. In addition,
one of our ambitious goals is to study cytochrome aa3 activity with NIRS. So far
we have not succeeded due to crosstalk. Nevertheless, the idea is of current interest
as it would tell us more about metabolism and be yet one more step forward in
resolving what forms the BOLD signal. Hopefully, with our fruitful international
collaboration we will be able to achieve more advanced signal separation of
metabolic substrates from the NIRS data.
Our aim is also to develop further our NIBP measurement method because it
may not always be as accurate as invasive BP measurement, which cannot be
recorded routinely. More advanced probe technology could increase the sensitivity
to beat-to-beat BP variability of the cardiac pulses, especially in subjects with a lot
of fat tissue that suppresses the propagation of pressure pulses. Moreover, poor
attachment of the acceleration sensor may result in an unclear signal response.
Therefore, improvements for the attachment method of the sensor, especially on
the neck, are still needed.
Finally, several diseases, such as ischemia, degenerative diseases,
inflammatory diseases, neoplasms and homeostatic disturbances have been linked
to the disruption of the protective properties of the BBB (Obermeier et al. 2013).
Recent advances in focused ultrasound (FUS) suggest that in the near future, BBBD
can be achieved less invasively in humans (O'Reilly & Hynynen 2012b). This
opens new possibilities for developing extended drug delivery in diseases affecting
either the BBB itself or the neuroglial tissue behind it (O'Reilly & Hynynen 2012a).
Due to our recent results, non-invasive NIRS and DC-EEG could be readily
coupled with such new methods for continuous monitoring of the treatment. Further
development of our setup for controlled drug delivery applications could also
benefit from the use of other in vivo optical imaging methods, such as Raman
spectroscopy (Wróbel et al. 2015a, Wróbel et al. 2015b) or optical
pharmacokinetics that provides tools for quantifying BBB penetrance of drugs in
animal models (Ergin et al. 2012). Furthermore, NIRS and DC-EEG are compatible
with ultrafast MRI, which could be used to obtain complementary information on
the brain status.
Actually, based on this, our ambitious goal is to build a new neuro-oncologic
Center here in Oulu, which will include, in addition to NIRS and DC-EEG, an
angiographic device, 3T MRI scanner and FUS in the immediate vicinity of each
other. With such a setup, the development of new biomarkers as well as diagnosis
and treatment of different oncologic patients, such as CNS lymphoma patients,
during BBBD will be easier in the future.
85
7 Conclusion
This doctoral dissertation showed that it is possible to measure brain dynamics from
the brain cortex using NIRS. The minimum source-detector distance in the NIRS
device should be at least 2.5 cm, but increasing the distance to 3 cm gives a better
signal. The comparison of different wavelength combinations indicated that the pair
830 nm and 660 nm gives the most accurate results. NIRS was also integrated to
be a part of a multimodal neuroimaging system, called Hepta-scan, in millisecond
synchrony. Our preliminary results revealed unknown non-stationarity between all
the multimodal signals measured. Furthermore, our results supported the general
agreement that the main source of LF BOLD oscillations in functionally connected
brain networks is electrophysiological activity coupled with neurovascular activity.
Lastly, combined NIRS and DC-EEG was able to demonstrate BBB opening during
treatment of human PCNSL. These observations demonstrate the feasibility of DC-
EEG and NIRS for real-time monitoring of BBB opening, showing that these
multimodal methods may be exploited when devising novel monitoring of BBB
integrity in brain diseases.
86
87
References
Aladjalova N (1957) Infra-slow rhythmic oscillations of the steady potential of the cerebral cortex. Nature 179: 957-959.
Alderliesten T, De Vis J, Lemmers P, Van Bel F, Benders M, Hendrikse J & Petersen E (2014) Simultaneous quantitative assessment of cerebral physiology using respiratory-calibrated MRI and near-infrared spectroscopy in healthy adults. Neuroimage 85: 255-263.
Allen PJ, Josephs O & Turner R (2000) A method for removing imaging artifact from continuous EEG recorded during functional MRI. Neuroimage 12(2): 230-239.
Allen PJ, Polizzi G, Krakow K, Fish DR & Lemieux L (1998) Identification of EEG events in the MR scanner: The problem of pulse artifact and a method for its subtraction. Neuroimage 8(3): 229-239.
Angelov L, Doolittle ND, Kraemer DF, Siegal T, Barnett GH, Peereboom DM, Stevens G, McGregor J, Jahnke K, Lacy CA, Hedrick NA, Shalom E, Ference S, Bell S, Sorenson L, Tyson RM, Haluska M & Neuwelt EA (2009) Blood-brain barrier disruption and intra-arterial methotrexate-based therapy for newly diagnosed primary CNS lymphoma: a multi-institutional experience. J Clin Oncol 27(21): 3503-3509.
Arai H, Takano M, Miyakawa K, Ota T, Takahashi T, Asaka H & Kawaguchi T (2006) A quantitative near-infrared spectroscopy study: a decrease in cerebral hemoglobin oxygenation in Alzheimer’s disease and mild cognitive impairment. Brain Cogn 61(2): 189-194.
Aries MJ, Coumou AD, Elting JW, van der Harst JJ, Kremer BP & Vroomen PC (2012) Near infrared spectroscopy for the detection of desaturations in vulnerable ischemic brain tissue: a pilot study at the stroke unit bedside. Stroke 43(4): 1134-1136.
Arridge SR, Cope M & Delpy D (1992) The theoretical basis for the determination of optical pathlengths in tissue: temporal and frequency analysis. Phys Med Biol 37(7): 1531-1560.
Ashburner J, Barnes G, Chen C, Daunizeau J, Flandin G, Friston K, Gitelman D, Kiebel S, Kilner J & Litvak V (2012) SPM8 manual. Functional Imaging Laboratory, Institute of Neurology, UCL: 1-475 .
Assländer J, Zahneisen B, Hugger T, Reisert M, Lee H-, LeVan P & Hennig J (2013) Single shot whole brain imaging using spherical stack of spirals trajectories. Neuroimage 73: 59-70.
Atsumori H, Kiguchi M, Katura T, Funane T, Obata A, Sato H, Manaka T, Iwamoto M, Maki A & Koizumi H (2010) Noninvasive imaging of prefrontal activation during attention-demanding tasks performed while walking using a wearable optical topography system. J Biomed Opt 15(4): 046002-1-046002-7.
Auer DP (2008) Spontaneous low-frequency blood oxygenation level-dependent fluctuations and functional connectivity analysis of the ‘resting’brain. Magn Reson Imaging 26(7): 1055-1064.
Bajaj S, Drake D, Butler AJ & Dhamala M (2014) Oscillatory motor network activity during rest and movement: an fNIRS study. Front.Syst.Neurosci 8(13): 1-12.
88
Beckmann CF, Mackay CE, Filippini N & Smith SM (2009) Group comparison of resting-state FMRI data using multi-subject ICA and dual regression. OHBM San Francisco, CA, USA: Poster #441.
Beckmann CF, Jenkinson M & Smith SM (2003) General multilevel linear modeling for group analysis in FMRI. Neuroimage 20(2): 1052-1063.
Beckmann CF, DeLuca M, Devlin JT & Smith SM (2005) Investigations into resting-state connectivity using independent component analysis. Phil Trans R Soc B 360(1457): 1001-1013.
Beckmann CF & Smith SM (2004) Probabilistic Independent Component Analysis for Functional Magnetic Resonance Imaging. IEEE Trans Med Imaging 23(2): 137-152.
Bell AJ & Sejnowski TJ (1995) An information-maximization approach to blind separation and blind deconvolution. Neural Comput 7(6): 1129-1159.
Berger H (1929) Über das elektrenkephalogramm des menschen. Eur Arch Psychiatry Clin Neurosci 87(1): 527-570.
Bevan PJ (2015) Should cerebral near-infrared spectroscopy be standard of care in adult cardiac surgery? Heart, Lung and Circulation 24(6): 544-550.
Bhambhani Y, Malik R & Mookerjee S (2007) Cerebral oxygenation declines at exercise intensities above the respiratory compensation threshold. Respiratory physiology & neurobiology 156(2): 196-202.
Bhattacharyya PK & Lowe MJ (2004) Cardiac-induced physiologic noise in tissue is a direct observation of cardiac-induced fluctuations. Magn Reson Imaging 22(1): 9-13.
Birn RM, Diamond JB, Smith MA & Bandettini PA (2006) Separating respiratory-variation-related fluctuations from neuronal-activity-related fluctuations in fMRI. Neuroimage 31(4): 1536-1548.
Birn RM, Murphy K & Bandettini PA (2008) The effect of respiration variations on independent component analysis results of resting state functional connectivity. Hum Brain Mapp 29(7): 740-750.
Biswal B, Zerrin Yetkin F, Haughton VM & Hyde JS (1995) Functional connectivity in the motor cortex of resting human brain using echo‐planar mri. Magnetic resonance in medicine 34(4): 537-541.
Biswal BB, Mennes M, Zuo XN, Gohel S, Kelly C, Smith SM, Beckmann CF, Adelstein JS, Buckner RL, Colcombe S, Dogonowski AM, Ernst M, Fair D, Hampson M, Hoptman MJ, Hyde JS, Kiviniemi VJ, Kotter R, Li SJ, Lin CP, Lowe MJ, Mackay C, Madden DJ, Madsen KH, Margulies DS, Mayberg HS, McMahon K, Monk CS, Mostofsky SH, Nagel BJ, Pekar JJ, Peltier SJ, Petersen SE, Riedl V, Rombouts SA, Rypma B, Schlaggar BL, Schmidt S, Seidler RD, Siegle GJ, Sorg C, Teng GJ, Veijola J, Villringer A, Walter M, Wang L, Weng XC, Whitfield-Gabrieli S, Williamson P, Windischberger C, Zang YF, Zhang HY, Castellanos FX & Milham MP (2010) Toward discovery science of human brain function. Proc Natl Acad Sci U S A 107(10): 4734-4739.
Boas DA, Gaudette T, Strangman G, Cheng X, Marota JJA & Mandeville JB (2001) The accuracy of near infrared spectroscopy and imaging during focal changes in cerebral hemodynamics. Neuroimage 13(1): 76-90.
89
Boly M, Garrido MI, Gosseries O, Bruno MA, Boveroux P, Schnakers C, Massimini M, Litvak V, Laureys S & Friston K (2011) Preserved feedforward but impaired top-down processes in the vegetative state. Science 332(6031): 858-862.
Bonnéry C, Leclerc P, Desjardins M, Hoge R, Bherer L, Pouliot P & Lesage F (2012) Changes in diffusion path length with old age in diffuse optical tomography. J Biomed Opt 17(5): 056002-1-056002-8.
Boyacioglu R & Barth M (2013) Generalized inverse imaging (GIN): Ultrafast fMRI with physiological noise correction. Magnetic Resonance in Medicine 70(4): 962-971.
Boyacioglu R, Beckmann CF & Barth M (2013) An investigation of RSN frequency spectra using ultra-fast generalized inverse imaging. Front Hum Neurosci 7(156): 1-7.
Bozkurt A & Onaral B (2004) Safety assessment of near infrared light emitting diodes for diffuse optical measurements. Biomed Eng Online 3(9): 1-10.
Buchheim K, Obrig H, Pannwitz Wv, Müller A, Heekeren H, Villringer A & Meierkord H (2004) Decrease in haemoglobin oxygenation during absence seizures in adult humans. Neurosci Lett 354(2): 119-122.
Budohoski KP, Czosnyka M, Smielewski P, Kasprowicz M, Helmy A, Bulters D, Pickard JD & Kirkpatrick PJ (2012) Impairment of cerebral autoregulation predicts delayed cerebral ischemia after subarachnoid hemorrhage: a prospective observational study. Stroke 43(12): 3230-3237.
Chance B, Zhuang Z, UnAh C, Alter C & Lipton L (1993) Cognition-activated low-frequency modulation of light absorption in human brain. Proc Natl Acad Sci U S A 90(8): 3770-3774.
Chang C, Cunningham JP & Glover GH (2009) Influence of heart rate on the BOLD signal: the cardiac response function. Neuroimage 44(3): 857-869.
Chang C & Glover GH (2009) Relationship between respiration, end-tidal CO 2, and BOLD signals in resting-state fMRI. Neuroimage 47(4): 1381-1393.
Chang C & Glover GH (2010) Time–frequency dynamics of resting-state brain connectivity measured with fMRI. Neuroimage 50(1): 81-98.
Chang C, Metzger CD, Glover GH, Duyn JH, Heinze H- & Walter M (2013) Association between heart rate variability and fluctuations in resting-state functional connectivity. Neuroimage 68: 93-104.
Chen L, Sandmann P, Thorne JD, Herrmann CS & Debener S (2015) Association of concurrent fNIRS and EEG signatures in response to auditory and visual stimuli. Brain Topogr 28(5): 710-725.
Cheong W, Prahl SA & Welch AJ (1990) A review of the optical properties of biological tissues. IEEE J Quant Electron 26(12): 2166-2185.
Chiarelli AM, Maclin EL, Fabiani M & Gratton G (2015) A kurtosis-based wavelet algorithm for motion artifact correction of fNIRS data. Neuroimage 112: 128-137.
Claassen JA, Levine BD & Zhang R (2009) Dynamic cerebral autoregulation during repeated squat-stand maneuvers. J Appl Physiol 106(1): 153-160.
Cole DM, Smith SM & Beckmann CF (2010) Advances and pitfalls in the analysis and interpretation of resting-state FMRI data. Front Syst Neurosci 4(8): 1-15.
90
Cooper RJ, Selb J, Gagnon L, Phillip D, Schytz HW, Iversen HK, Ashina M & Boas DA (2012a) A systematic comparison of motion artifact correction techniques for functional near-infrared spectroscopy. Front Neurosci 6(147): 1-10.
Cooper RJ, Gagnon L, Goldenholz DM, Boas DA & Greve DN (2012b) The utility of near-infrared spectroscopy in the regression of low-frequency physiological noise from functional magnetic resonance imaging data. Neuroimage 59(4): 3128-3138.
Cope M (1991) The application of near infrared spectroscopy to non invasive monitoring of cerebral oxygenation in the newborn infant. PhD thesis. University College London, Department of Medical Physics and Bioengineering.
Cope M & Delpy D (1988) System for long-term measurement of cerebral blood and tissue oxygenation on newborn infants by near infra-red transillumination. Medical and Biological Engineering and Computing 26(3): 289-294.
Cope M, Delpy D, Reynolds E, Wray S, Wyatt J & Van der Zee P (1988) Methods of quantitating cerebral near infrared spectroscopy data. In: Mochizuki M, Honig CR, Koyama T, Goldstick TK & Bruley DF (eds) Oxygen Transport to Tissue X. New York, Springer US: 183-189.
Corlu A, Choe R, Durduran T, Lee K, Schweiger M, Arridge SR, Hillman EM & Yodh AG (2005) Diffuse optical tomography with spectral constraints and wavelength optimization. Appl Opt 44(11): 2082-2093.
Correia T, Gibson A & Hebden J (2010) Identification of the optimal wavelengths for optical topography: a photon measurement density function analysis. J Biomed Opt 15(5): 056002-1-056002-14.
Cox RW (1996) AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Computers and Biomedical research 29(3): 162-173.
Cox RW (2012) AFNI: what a long strange trip it's been. Neuroimage 62(2): 743-747. Coyle SM, Ward TE & Markham CM (2007) Brain–computer interface using a simplified
functional near-infrared spectroscopy system. Journal of neural engineering 4(3): 219-226.
Cui X, Bray S, Bryant DM, Glover GH & Reiss AL (2011) A quantitative comparison of NIRS and fMRI across multiple cognitive tasks. Neuroimage 54(4): 2808-2821.
Dagli MS, Ingeholm JE & Haxby JV (1999) Localization of cardiac-induced signal change in fMRI. Neuroimage 9(4): 407-415.
Delorme A, Mullen T, Kothe C, Acar ZA, Bigdely-Shamlo N, Vankov A & Makeig S (2011) EEGLAB, SIFT, NFT, BCILAB, and ERICA: new tools for advanced EEG processing. Computational intelligence and neuroscience 2011(130714): 1-12.
Delorme A & Makeig S (2004) EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods 134(1): 9-21.
Delpy DT, Cope M, van der Zee P, Arridge S, Wray S & Wyatt J (1988) Estimation of optical pathlength through tissue from direct time of flight measurement. Phys Med Biol 33(12): 1433-1442.
Ding XP, Gao X, Fu G & Lee K (2013) Neural correlates of spontaneous deception: A functional near-infrared spectroscopy (fNIRS) study. Neuropsychologia 51(4): 704-712.
91
Doolittle ND, Abrey LE, Shenkier TN, Tali S, Bromberg JE, Neuwelt EA, Soussain C, Jahnke K, Johnston P, Illerhaus G, Schiff D, Batchelor T, Montoto S, Kraemer DF & Zucca E (2008) Brain parenchyma involvement as isolated central nervous system relapse of systemic non-Hodgkin lymphoma: an International Primary CNS Lymphoma Collaborative Group report. Blood 111(3): 1085-1093.
Doolittle ND, Anderson CP, Bleyer WA, Cairncross JG, Cloughesy T, Eck SL, Guastadisegni P, Hall WA, Muldoon LL, Patel SJ, Peereboom D, Siegal T & Neuwelt EA (2001) Importance of dose intensity in neuro-oncology clinical trials: summary report of the Sixth Annual Meeting of the Blood-Brain Barrier Disruption Consortium. Neuro Oncol 3(1): 46-54.
Doolittle ND, Muldoon LL, Culp AY & Neuwelt EA (2014) Delivery of chemotherapeutics across the blood-brain barrier: Challenges and advances. Advances in Pharmacology 71: 203-243.
Duan L, Zhang Y- & Zhu C- (2012) Quantitative comparison of resting-state functional connectivity derived from fNIRS and fMRI: A simultaneous recording study. Neuroimage 60(4): 2008-2018.
Duncan A, Meek JH, Clemence M, Elwell CE, Fallon P, Tyszczuk L, Cope M & Delpy DT (1996) Measurement of cranial optical path length as a function of age using phase resolved near infrared spectroscopy. Pediatr Res 39(5): 889-894.
Duncan A, Meek JH, Clemence M, Elwell CE, Tyszczuk L, Cope M & Delpy D (1995) Optical pathlength measurements on adult head, calf and forearm and the head of the newborn infant using phase resolved optical spectroscopy. Phys Med Biol 40(2): 295-304.
Ergin A, Wang M, Zhang J, Bigio I & Joshi S (2012) Noninvasive in vivo optical assessment of blood brain barrier permeability and brain tissue drug deposition in rabbits. J Biomed Opt 17(5): 057008-1-057008-10.
Essenpreis M, Elwell C, Cope M, Van der Zee P, Arridge S & Delpy D (1993) Spectral dependence of temporal point spread functions in human tissues. Appl Opt 32(4): 418-425.
Fazli S, Mehnert J, Steinbrink J, Curio G, Villringer A, Müller K & Blankertz B (2012) Enhanced performance by a hybrid NIRS–EEG brain computer interface. Neuroimage 59(1): 519-529.
Fekete T, Rubin D, Carlson JM & Mujica-Parodi LR (2011a) A stand-alone method for anatomical localization of NIRS measurements. Neuroimage 56(4): 2080-2088.
Fekete T, Rubin D, Carlson JM & Mujica-Parodi LR (2011b) The NIRS Analysis Package: noise reduction and statistical inference. PLoS One 6(9): e24322, 1-21.
Ferrari M & Quaresima V (2012) A brief review on the history of human functional near-infrared spectroscopy (fNIRS) development and fields of application. Neuroimage 63(2): 921-935.
Ferrari M, Wei Q, De Blasi RA, Quaresima V & Zaccanti G (1993) Variability of human brain and muscle optical pathlength in different experimental conditions. Proc. SPIE1888, Photon Migration and Imaging in Random Media and Tissues. : 466-472.
92
Ferreri L, Bigand E, Perrey S & Bugaiska A (2014) The promise of Near-Infrared Spectroscopy (NIRS) for psychological research: a brief review. L’Année psychologique 114(3): 537-569.
Filippini N, MacIntosh BJ, Hough MG, Goodwin GM, Frisoni GB, Smith SM, Matthews PM, Beckmann CF & Mackay CE (2009) Distinct patterns of brain activity in young carriers of the APOE-e4 allele. Proc Natl Acad Sci U S A 106(17): 7209-7214.
Flodin P (2015) Intrinsic brain activity in health and disease. PhD thesis. Karolinska Institutet, Department of Clinical Neuroscience.
Fox MD & Greicius M (2010) Clinical applications of resting state functional connectivity. Front Syst Neurosci 4(19): 1-13.
Fox MD & Raichle ME (2007) Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nat Rev Neurosci 8(9): 700-711.
Fox PT & Raichle ME (1986) Focal physiological uncoupling of cerebral blood flow and oxidative metabolism during somatosensory stimulation in human subjects. Proc Natl Acad Sci U S A 83(4): 1140-1144.
Friston KJ (2011) Functional and effective connectivity: a review. Brain Connectivity 1(1): 13-36.
Friston KJ, Holmes AP, Worsley KJ, Poline J, Frith CD & Frackowiak RS (1994) Statistical parametric maps in functional imaging: a general linear approach. Hum Brain Mapp 2(4): 189-210.
Fujiwara N, Sakatani K, Katayama Y, Murata Y, Hoshino T, Fukaya C & Yamamoto T (2004) Evoked-cerebral blood oxygenation changes in false-negative activations in BOLD contrast functional MRI of patients with brain tumors. Neuroimage 21(4): 1464-1471.
Funane T, Atsumori H, Sato H, Kiguchi M & Maki A (2009) Relationship between wavelength combination and signal-to-noise ratio in measuring hemoglobin concentrations using visible or near-infrared light. Optical review 16(4): 442-448.
Furusho J, Suzuki A, Takakusa Y, Kawaguchi F, Ichikawa N & Kato T (2002) Simultaneous study of interictal EEG and near-infrared spectroscopy in a boy with epilepsy. International Congress Series. 1232: 673-676.
Gagnon L, Yücel MA, Dehaes M, Cooper RJ, Perdue KL, Selb J, Huppert TJ, Hoge RD & Boas DA (2012) Quantification of the cortical contribution to the NIRS signal over the motor cortex using concurrent NIRS-fMRI measurements. Neuroimage 59(4): 3933-3940.
Gallagher A, Lassonde M, Bastien D, Vannasing P, Lesage F, Grova C, Bouthillier A, Carmant L, Lepore F & Béland R (2008) Non-invasive pre-surgical investigation of a 10 year-old epileptic boy using simultaneous EEG–NIRS. Seizure 17(6): 576-582.
Germon TJ, Evans PD, Barnett NJ, Wall P, Manara AR & Nelson RJ (1999) Cerebral near infrared spectroscopy: emitter-detector separation must be increased. Br J Anaesth 82(6): 831-837.
Gesche H, Grosskurth D, Küchler G & Patzak A (2012) Continuous blood pressure measurement by using the pulse transit time: Comparison to a cuff-based method. Eur J Appl Physiol 112(1): 309-315.
93
Gibson A, Hebden J & Arridge SR (2005) Recent advances in diffuse optical imaging. Phys Med Biol 50(4): R1-R43.
Gorshkov AV & Kirillin MY (2012) Monte Carlo simulation of brain sensing by optical diffuse spectroscopy. Journal of Computational Science 3(6): 498-503.
Gratton E, Toronov V, Wolf U, Wolf M & Webb A (2005) Measurement of brain activity by near-infrared light. J Biomed Opt 10(1): 011008-1-011008-13.
Greicius MD, Krasnow B, Reiss AL & Menon V (2003) Functional connectivity in the resting brain: a network analysis of the default mode hypothesis. Proc Natl Acad Sci U S A 100(1): 253-258.
Gribbin B, Steptoe A & Sleight P (1976) Pulse wave velocity as a measure of blood pressure change. Psychophysiology 13(1): 86-90.
Haginoya K, Munakata M, Kato R, Yokoyama H, Ishizuka M & Iinuma K (2002) Ictal cerebral haemodynamics of childhood epilepsy measured with near-infrared spectrophotometry. Brain 125(Pt 9): 1960-1971.
Heekeren HR, Kohl M, Obrig H, Wenzel R, von Pannwitz W, Matcher SJ, Dirnagl U, Cooper CE & Villringer A (1999) Noninvasive assessment of changes in cytochrome-c oxidase oxidation in human subjects during visual stimulation. Journal of Cerebral Blood Flow & Metabolism 19(6): 592-603.
HELD D, FENCL V & PAPPENHEIMER JR (1964) Electrical Potential of Cerebrospinal Fluid. J Neurophysiol 27: 942-959.
Hennig J, Zhong K & Speck O (2007) MR-Encephalography: Fast multi-channel monitoring of brain physiology with magnetic resonance. Neuroimage 34(1): 212-219.
Hiltunen T, Kantola J, Elseoud AA, Lepola P, Suominen K, Starck T, Nikkinen J, Remes J, Tervonen O, Palva S, Kiviniemi V & Matias Palva J (2014) Infra-slow EEG fluctuations are correlated with resting-state network dynamics in fMRI. Journal of Neuroscience 34(2): 356-362.
Holper L, Scholkmann F & Wolf M (2014) The relationship between sympathetic nervous activity and cerebral hemodynamics and oxygenation: A study using skin conductance measurement and functional near-infrared spectroscopy. Behav Brain Res 270: 95-107.
Hong K, Naseer N & Kim Y (2015) Classification of prefrontal and motor cortex signals for three-class fNIRS–BCI. Neurosci Lett 587: 87-92.
Hoshi Y, Kosaka S, Xie Y, Kohri S & Tamura M (1998) Relationship between fluctuations in the cerebral hemoglobin oxygenation state and neuronal activity under resting conditions in man. Neurosci Lett 245(3): 147-150.
Hoshi Y, Kobayashi N & Tamura M (2001) Interpretation of near-infrared spectroscopy signals: a study with a newly developed perfused rat brain model. J Appl Physiol 90(5): 1657-1662.
Hoshi Y (2003) Functional near-infrared optical imaging: Utility and limitations in human brain mapping. Psychophysiology 40(4): 511-520.
Huppert TJ, Diamond SG, Franceschini MA & Boas DA (2009) HomER: a review of time-series analysis methods for near-infrared spectroscopy of the brain. Appl Opt 48(10): D280-D298.
94
Huppert TJ, Hoge RD, Diamond SG, Franceschini MA & Boas DA (2006) A temporal comparison of BOLD, ASL, and NIRS hemodynamic responses to motor stimuli in adult humans. Neuroimage 29(2): 368-382.
Huster RJ, Debener S, Eichele T & Herrmann CS (2012) Methods for simultaneous EEG-fMRI: an introductory review. J Neurosci 32(18): 6053-6060.
Hutchison RM, Womelsdorf T, Allen EA, Bandettini PA, Calhoun VD, Corbetta M, Della Penna S, Duyn JH, Glover GH, Gonzalez-Castillo J, Handwerker DA, Keilholz S, Kiviniemi V, Leopold DA, de Pasquale F, Sporns O, Walter M & Chang C (2013) Dynamic functional connectivity: Promise, issues, and interpretations. Neuroimage 80: 360-378.
Indic P, Salisbury EB, Paydarfar D, Brown EN & Barbieri R (2008) Interaction between heart rate variability and respiration in preterm infants. Comput Cardiol 35: 57-60.
Irani F, Platek SM, Bunce S, Ruocco AC & Chute D (2007) Functional near infrared spectroscopy (fNIRS): an emerging neuroimaging technology with important applications for the study of brain disorders. Clin Neuropsychol 21(1): 9-37.
Jackson PA & Kennedy DO (2013) The application of near infrared spectroscopy in nutritional intervention studies. Front Hum Neurosci 7(473): 1-6.
Jang KE, Tak S, Jung J, Jang J, Jeong Y & Ye JC (2009) Wavelet minimum description length detrending for near-infrared spectroscopy. J Biomed Opt 14(3): 034004-1-034004-13.
Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW & Smith SM (2012) Fsl. Neuroimage 62(2): 782-790.
Jenkinson M, Bannister P, Brady M & Smith S (2002) Improved optimization for the robust and accurate linear registration and motion correction of brain images. Neuroimage 17(2): 825-841.
Jobsis FF (1977) Noninvasive, infrared monitoring of cerebral and myocardial oxygen sufficiency and circulatory parameters. Science 198(4323): 1264-1267.
Julien C (2006) The enigma of Mayer waves: Facts and models. Cardiovasc Res 70(1): 12-21.
Kasenda B, Ferreri AJ, Marturano E, Forst D, Bromberg J, Ghesquieres H, Ferlay C, Blay JY, Hoang-Xuan K, Pulczynski EJ, Fossa A, Okoshi Y, Chiba S, Fritsch K, Omuro A, O'Neill BP, Bairey O, Schandelmaier S, Gloy V, Bhatnagar N, Haug S, Rahner S, Batchelor TT, Illerhaus G & Briel M (2015) First-line treatment and outcome of elderly patients with primary central nervous system lymphoma (PCNSL)-a systematic review and individual patient data meta-analysis. Ann Oncol 26(7): 1305-1313.
Kawaguchi H, Okui N, Sakaguchi K & Okada E (2008) Theoretical analysis of crosstalk between oxygenated and deoxygenated haemoglobin in focal brain-activation measurements by near-infrared topography. Opto-Electronics Review 16(4): 404-412.
Khan MJ, Hong MJ & Hong K (2014) Decoding of four movement directions using hybrid NIRS-EEG brain-computer interface. Front Hum Neurosci 8(244): 1-10.
Kirilina E, Yu N, Jelzow A, Wabnitz H, Jacobs AM & Tachtsidis I (2013) Identifying and quantifying main components of physiological noise in functional near infrared spectroscopy on the prefrontal cortex. Front Hum Neurosci 7(864): 1-17.
95
Kiviniemi AM, Tiinanen S, Hautala AJ, Seppänen T, Mäkikallio TH, Huikuri HV & Tulppo MP (2010) Frequency of slow oscillations in arterial pressure and R–R intervals during muscle metaboreflex activation. Autonomic Neuroscience 152(1): 88-95.
Kiviniemi V (2008) Endogenous brain fluctuations and diagnostic imaging. Hum Brain Mapp 29(7): 810-817.
Kiviniemi V, Jauhiainen J, Tervonen O, Pääkkö E, Oikarinen J, Vainionpää V, Rantala H & Biswal B (2000) Slow vasomotor fluctuation in fMRI of anesthetized child brain. Magnetic Resonance in Medicine 44(3): 373-378.
Kiviniemi V, Kantola J-, Jauhiainen J, Hyvärinen A & Tervonen O (2003) Independent component analysis of nondeterministic fMRI signal sources. Neuroimage 19(2): 253-260.
Kiviniemi V, Ruohonen J & Tervonen O (2005) Separation of physiological very low frequency fluctuation from aliasing by switched sampling interval fMRI scans. Magn Reson Imaging 23(1): 41-46.
Kiviniemi V, Starck T, Remes J, Long X, Nikkinen J, Haapea M, Veijola J, Moilanen I, Isohanni M, Zang Y- & Tervonen O (2009) Functional segmentation of the brain cortex using high model order group PICA. Hum Brain Mapp 30(12): 3865-3886.
Kiviniemi V, Wang X, Korhonen V, Keinanen T, Tuovinen T, Autio J, LeVan P, Keilholz S, Zang YF, Hennig J & Nedergaard M (2016) Ultra-fast magnetic resonance encephalography of physiological brain activity - Glymphatic pulsation mechanisms? J Cereb Blood Flow Metab 36(6): 1033-1045.
Kleinschmidt A, Obrig H, Requardt M, Merboldt K, Dirnagl U, Villringer A & Frahm J (1996) Simultaneous recording of cerebral blood oxygenation changes during human brain activation by magnetic resonance imaging and near-infrared spectroscopy. Journal of cerebral blood flow & metabolism 16(5): 817-826.
Koch SP, Koendgen S, Bourayou R, Steinbrink J & Obrig H (2008) Individual alpha-frequency correlates with amplitude of visual evoked potential and hemodynamic response. Neuroimage 41(2): 233-242.
Koo B, Lee H, Nam Y, Kang H, Koh CS, Shin H & Choi S (2015) A hybrid NIRS-EEG system for self-paced brain computer interface with online motor imagery. J Neurosci Methods 244: 26-32.
Kraemer DF, Fortin D, Doolittle ND & Neuwelt EA (2001) Association of total dose intensity of chemotherapy in primary central nervous system lymphoma (human non-acquired immunodeficiency syndrome) and survival. Neurosurgery 48(5): 1033-1041.
Kropotov J (2010) Quantitative EEG, event-related potentials and neurotherapy. San Diego, CA, Academic Press.
Lareau E, Lesage F, Pouliot P, Nguyen D, Le Lan J & Sawan M (2011) Multichannel wearable system dedicated for simultaneous electroencephalography/near-infrared spectroscopy real-time data acquisitions. J Biomed Opt 16(9): 096014-1-096014-14.
Laufs H (2012) A personalized history of EEG–fMRI integration. Neuroimage 62(2): 1056-1067.
Lauterbur PC (1973) Image formation by induced local interactions: examples employing nuclear magnetic resonance. Nature 242(5394): 190-191.
96
Lee M, Fazli S, Mehnert J & Lee S (2015) Subject-dependent classification for robust idle state detection using multi-modal neuroimaging and data-fusion techniques in BCI. Pattern Recognit 48(8): 2725-2737.
Lee H-, Zahneisen B, Hugger T, LeVan P & Hennig J (2013) Tracking dynamic resting-state networks at higher frequencies using MR-encephalography. Neuroimage 65: 216-222.
Li H, Tak S & Ye JC (2012) Lipschitz-Killing curvature based expected Euler characteristics for p-value correction in fNIRS. J Neurosci Methods 204(1): 61-67.
Liboni W, Molinari F, Allais G, Mana O, Negri E, Grippi G, Benedetto C, D'Andrea G & Bussone G (2007) Why do we need NIRS in migraine? Neurological Sciences 28(2): S222-S224.
Lin F-, Tsai KWK, Chu Y-, Witzel T, Nummenmaa A, Raij T, Ahveninen J, Kuo W- & Belliveau JW (2012) Ultrafast inverse imaging techniques for fMRI. Neuroimage 62(2): 699-705.
Lloyd-Fox S, Blasi A & Elwell C (2010) Illuminating the developing brain: the past, present and future of functional near infrared spectroscopy. Neuroscience & Biobehavioral Reviews 34(3): 269-284.
Lu C, Zhang Y, Biswal BB, Zang Y, Peng D & Zhu C (2010) Use of fNIRS to assess resting state functional connectivity. J Neurosci Methods 186(2): 242-249.
Machado A, Lina J, Tremblay J, Lassonde M, Nguyen DK, Lesage F & Grova C (2011) Detection of hemodynamic responses to epileptic activity using simultaneous Electro-EncephaloGraphy (EEG)/Near Infra Red Spectroscopy (NIRS) acquisitions. Neuroimage 56(1): 114-125.
MacIntosh BJ, Klassen LM & Menon RS (2003) Transient hemodynamics during a breath hold challenge in a two part functional imaging study with simultaneous near-infrared spectroscopy in adult humans. Neuroimage 20(2): 1246-1252.
Maggioni E, Molteni E, Zucca C, Reni G, Cerutti S, Triulzi FM, Arrigoni F & Bianchi AM (2015) Investigation of negative BOLD responses in human brain through NIRS technique. A visual stimulation study. Neuroimage 108: 410-422.
Majeed W, Magnuson M, Hasenkamp W, Schwarb H, Schumacher EH, Barsalou L & Keilholz SD (2011a) Spatiotemporal dynamics of low frequency BOLD fluctuations in rats and humans. Neuroimage 54(2): 1140-1150.
Majeed W, Magnuson M, Hasenkamp W, Schwarb H, Schumacher EH, Barsalou L & Keilholz SD (2011b) Spatiotemporal dynamics of low frequency BOLD fluctuations in rats and humans. Neuroimage 54(2): 1140-1150.
Maki A, Yamashita Y, Ito Y, Watanabe E, Mayanagi Y & Koizumi H (1995) Spatial and temporal analysis of human motor activity using noninvasive NIR topography. Med Phys 22(12): 1997-2005.
Mansfield P (1977) Multi-planar image formation using NMR spin echoes. Journal of Physics C: Solid State Physics 10(3): L55-L58.
Mayer S (1876) Studien zur Physiologie des Herzens und der Blutgefässe 6. Abhandlung: Über spontane Blutdruckschwankungen. Sitzungsberichte Akademie der Wissenschaften in Wien.Mathematisch-naturwissenschaftliche Classe, Anatomie 74: 281-307.
97
McCormick PW, Stewart M, Lewis G, Dujovny M & Ausman JI (1992) Intracerebral penetration of infrared light: technical note. J Neurosurg 76(2): 315-318.
Medvedev AV (2014) Does the resting state connectivity have hemispheric asymmetry? A near-infrared spectroscopy study. Neuroimage 85: 400-407.
Mesquita RC, Franceschini MA & Boas DA (2010) Resting state functional connectivity of the whole head with near-infrared spectroscopy. Biomedical Optics Express 1(1): 324-336.
Minati L, Kress IU, Visani E, Medford N & Critchley HD (2011) Intra-and extra-cranial effects of transient blood pressure changes on brain near-infrared spectroscopy (NIRS) measurements. J Neurosci Methods 197(2): 283-288.
Miyai I, Tanabe HC, Sase I, Eda H, Oda I, Konishi I, Tsunazawa Y, Suzuki T, Yanagida T & Kubota K (2001) Cortical Mapping of Gait in Humans: A Near-Infrared Spectroscopic Topography Study. Neuroimage 14(5): 1186-1192.
Moerman A, Meert F & De Hert S (2015) Cerebral near-infrared spectroscopy in the care of patients during cardiological procedures: a summary of the clinical evidence. J Clin Monit Comput : 1-9.
Molinari F, Liboni W, Grippi G & Negri E (2006) Relationship between oxygen supply and cerebral blood flow assessed by transcranial Doppler and near–infrared spectroscopy in healthy subjects during breath–holding. Journal of neuroengineering and rehabilitation 3(16): 1-13.
Monto S, Palva S, Voipio J & Palva JM (2008) Very slow EEG fluctuations predict the dynamics of stimulus detection and oscillation amplitudes in humans. J Neurosci 28(33): 8268-8272.
Moosmann M, Ritter P, Krastel I, Brink A, Thees S, Blankenburg F, Taskin B, Obrig H & Villringer A (2003) Correlates of alpha rhythm in functional magnetic resonance imaging and near infrared spectroscopy. Neuroimage 20(1): 145-158.
Murkin JM & Arango M (2009) Near-infrared spectroscopy as an index of brain and tissue oxygenation. British Journal of Anaesthesia 103(suppl 1): i3-i13.
Myllylä T (2014) Multimodal biomedical measurement methods to study brain functions simultaneously with functional magnetic resonance imaging. PhD thesis. University of Oulu, Department of Electrical Engineering.
Myllylä TS, Vihriälä EV, Korhonen VO & Sorvoja HS (2013a) Case study of ECG signal used as a reference signal in optical pulse transit time measurement of blood flow: the effect of different electrode placements on pulse transit time. Proc. SPIE8699, Saratov Fall Meeting 2012: Laser Physics and Photonics XIV. paper 869903: 1-13.
Myllylä T, Toronov V, Claassen J, Kiviniemi V & Tuchin V (2016) The use of near-infrared spectroscopy in multimodal brain research. In: Tuchin V (ed) Handbook of Optical Biomedical Diagnostics, 2nd edition. Bellingham, Washington, SPIE press: 687-733.
Myllylä T, Korhonen V, Vihriälä E, Sorvoja H, Hiltunen T, Tervonen O & Kiviniemi V (2012) Human heart pulse wave responses measured simultaneously at several sensor placements by two MR-compatible fibre optic methods. J Sensors 2012(769613): 1-8.
98
Myllylä T, Popov A, Korhonen V, Bykov A & Kinnunen M (2013b) Optical sensing of a pulsating liquid in a brain-mimicking phantom. Proc. SPIE8799, Diffuse Optical Imaging IV. paper 87990X: 1-7.
Myllylä TS, Elseoud AA, Sorvoja HSS, Myllylä RA, Harja JM, Nikkinen J, Tervonen O & Kiviniemi V (2011) Fibre optic sensor for non-invasive monitoring of blood pressure during MRI scanning. Journal of Biophotonics 4(1-2): 98-107.
Näsi T, Virtanen J, Noponen T, Toppila J, Salmi T & Ilmoniemi RJ (2011) Spontaneous hemodynamic oscillations during human sleep and sleep stage transitions characterized with near-infrared spectroscopy. PLoS One 6(10): e25415, 1-9.
Neuwelt EA, Bauer B, Fahlke C, Fricker G, Iadecola C, Janigro D, Leybaert L, Molnár Z, O'Donnell ME & Povlishock JT (2011) Engaging neuroscience to advance translational research in brain barrier biology. Nature Reviews Neuroscience 12(3): 169-182.
Neuwelt EA, Frenkel EP, Diehl JT, Maravilla KR, Vu LH, Clark WK, Rapoport SI, Barnett PA, Hill SA, Lewis SE, Ehle AL, Beyer CW,Jr & Moore RJ (1979a) Osmotic blood-brain barrier disruption: a new means of increasing chemotherapeutic agent delivery. Trans Am Neurol Assoc 104: 256-260.
Neuwelt EA, Maravilla KR, Frenkel EP, Rapaport SI, Hill SA & Barnett PA (1979b) Osmotic blood-brain barrier disruption. Computerized tomographic monitoring of chemotherapeutic agent delivery. J Clin Invest 64(2): 684-688.
Nguyen DK, Tremblay J, Pouliot P, Vannasing P, Florea O, Carmant L, Lepore F, Sawan M, Lesage F & Lassonde M (2012) Non-invasive continuous EEG-fNIRS recording of temporal lobe seizures. Epilepsy Res 99(1): 112-126.
Nguyen DK, Tremblay J, Pouliot P, Vannasing P, Florea O, Carmant L, Lepore F, Sawan M, Lesage F & Lassonde M (2013) Noninvasive continuous functional near‐infrared spectroscopy combined with electroencephalography recording of frontal lobe seizures. Epilepsia 54(2): 331-340.
Niedermeyer E & da Silva FL (2005) Electroencephalography: basic principles, clinical applications, and related fields. Philadelphia, Lippincott Williams & Wilkins.
Nita DA, Vanhatalo S, Lafortune FD, Voipio J, Kaila K & Amzica F (2004) Nonneuronal origin of CO2-related DC EEG shifts: an in vivo study in the cat. J Neurophysiol 92(2): 1011-1022.
Obermeier B, Daneman R & Ransohoff RM (2013) Development, maintenance and disruption of the blood-brain barrier. Nat Med 19(12): 1584-1596.
Obrig H (2014) NIRS in clinical neurology—a ‘promising’tool? Neuroimage 85: 535-546. Obrig H, Hirth C, Junge-Hulsing JG, Doge C, Wolf T, Dirnagl U & Villringer A (1996)
Cerebral oxygenation changes in response to motor stimulation. J Appl Physiol (1985) 81(3): 1174-1183.
Obrig H & Steinbrink J (2011) Non-invasive optical imaging of stroke. Philos Trans A Math Phys Eng Sci 369(1955): 4470-4494.
Obrig H, Neufang M, Wenzel R, Kohl M, Steinbrink J, Einhäupl K & Villringer A (2000) Spontaneous Low Frequency Oscillations of Cerebral Hemodynamics and Metabolism in Human Adults. Neuroimage 12(6): 623-639.
99
Ogawa Y, Kotani K & Jimbo Y (2014) Relationship between working memory performance and neural activation measured using near‐infrared spectroscopy. Brain and behavior 4(4): 544-551.
Ogawa S, Lee TM, Kay AR & Tank DW (1990) Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc Natl Acad Sci U S A 87(24): 9868-9872.
Ogawa S, Tank DW, Menon R, Ellermann JM, Kim S-, Merkle H & Ugurbil K (1992) Intrinsic signal changes accompanying sensory stimulation: Functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci U S A 89(13): 5951-5955.
Okada E & Delpy DT (2003) Near-infrared light propagation in an adult head model. II. Effect of superficial tissue thickness on the sensitivity of the near-infrared spectroscopy signal. Appl Opt 42(16): 2915-2921.
Okada E, Firbank M, Schweiger M, Arridge SR, Cope M & Delpy DT (1997) Theoretical and experimental investigation of near-infrared light propagation in a model of the adult head. Appl Opt 36(1): 21-31.
Okui N & Okada E (2005) Wavelength dependence of crosstalk in dual-wavelength measurement of oxy-and deoxy-hemoglobin. J Biomed Opt 10(1): 011015-1-011015-8.
Oldag A, Goertler M, Bertz AK, Schreiber S, Stoppel C, Heinze HJ & Kopitzki K (2012) Assessment of cortical hemodynamics by multichannel near-infrared spectroscopy in steno-occlusive disease of the middle cerebral artery. Stroke 43(11): 2980-2985.
O'Reilly MA & Hynynen K (2012a) Ultrasound enhanced drug delivery to the brain and central nervous system. International Journal of Hyperthermia 28(4): 386-396.
O'Reilly M & Hynynen K (2012b) Blood-Brain Barrier: Real-time feedback-controlled focused ultrasound disruption by using an acoustic emissions–based controller. Radiology 263(1): 96-106.
Park H, Correia S, Ducorps A & Tallon-Baudry C (2014) Spontaneous fluctuations in neural responses to heartbeats predict visual detection. Nat Neurosci 17(4): 612-618.
Peng K, Nguyen DK, Tayah T, Vannasing P, Tremblay J, Sawan M, Lassonde M, Lesage F & Pouliot P (2014) fNIRS-EEG study of focal interictal epileptiform discharges. Epilepsy Res 108(3): 491-505.
Pfurtscheller G, Bauernfeind G, Neuper C & da Silva, Fernando H Lopes (2012a) Does conscious intention to perform a motor act depend on slow prefrontal (de) oxyhemoglobin oscillations in the resting brain? Neurosci Lett 508(2): 89-94.
Pfurtscheller G, Bauernfeind G, Wriessnegger SC & Neuper C (2010a) Focal frontal (de) oxyhemoglobin responses during simple arithmetic. International Journal of Psychophysiology 76(3): 186-192.
Pfurtscheller G, Klobassa DS, Altstätter C, Bauernfeind G & Neuper C (2011) About the stability of phase shifts between slow oscillations around 0.1 Hz in cardiovascular and cerebral systems. Biomedical Engineering, IEEE Transactions on 58(7): 2064-2071.
Pfurtscheller G, Ortner R, Bauernfeind G, Linortner P & Neuper C (2010b) Does conscious intention to perform a motor act depend on slow cardiovascular rhythms? Neurosci Lett 468(1): 46-50.
100
Pfurtscheller G, Allison BZ, Brunner C, Bauernfeind G, Solis-Escalante T, Scherer R, Zander TO, Mueller-Putz G, Neuper C & Birbaumer N (2010c) The hybrid BCI. Front Neurosci 4(42): 1-11.
Pfurtscheller G, Daly I, Bauernfeind G & Muller-Putz GR (2012b) Coupling between intrinsic prefrontal HbO2 and central EEG beta power oscillations in the resting brain. PLoS One 7(8): e43640.
Pinti P, Cardone D & Merla A (2015) Simultaneous fNIRS and thermal infrared imaging during cognitive task reveal autonomic correlates of prefrontal cortex activity. Sci Rep 5(17471): 1-14.
Piper SK, Krueger A, Koch SP, Mehnert J, Habermehl C, Steinbrink J, Obrig H & Schmitz CH (2014) A wearable multi-channel fNIRS system for brain imaging in freely moving subjects. Neuroimage 85: 64-71.
Pizza F, Biallas M, Kallweit U, Wolf M & Bassetti CL (2012) Cerebral hemodynamic changes in stroke during sleep-disordered breathing. Stroke 43(7): 1951-1953.
Posse S, Ackley E, Mutihac R, Zhang T, Hummatov R, Akhtari M, Chohan M, Fisch B & Yonas H (2013) High-speed real-time resting-state FMRI using multi-slab echo-volumar imaging. Front Hum Neurosci 7(479): 1-23.
Pouliot P, Tran TPY, Birca V, Vannasing P, Tremblay J, Lassonde M & Nguyen DK (2014) Hemodynamic changes during posterior epilepsies: An EEG-fNIRS study. Epilepsy Res 108(5): 883-890.
Pouliot P, Tremblay J, Robert M, Vannasing P, Lepore F, Lassonde M, Sawan M, Nguyen DK & Lesage F (2012) Nonlinear hemodynamic responses in human epilepsy: a multimodal analysis with fNIRS-EEG and fMRI-EEG. J Neurosci Methods 204(2): 326-340.
Power SD, Kushki A & Chau T (2012) Automatic single-trial discrimination of mental arithmetic, mental singing and the no-control state from prefrontal activity: toward a three-state NIRS-BCI. BMC Res Notes 5(141): 1-10.
Prahl S (1999) Optical Absorption of Hemoglobin. URI: http://omlc.org/spectra/hemoglobin/index.html. Cited 25.08.2015.
Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA & Shulman GL (2001) A default mode of brain function. Proc Natl Acad Sci U S A 98(2): 676-682.
Revest PA, Jones HC & Abbott NJ (1993) The transendothelial DC potential of rat blood-brain barrier vessels in situ. In: Drewes LR & Betz AL (eds) Frontiers in Cerebral Vascular Biology. New York, Springer US: 71-74.
Revest PA, Jones HC & Abbott NJ (1994) Transendothelial electrical potential across pial vessels in anaesthetised rats: a study of ion permeability and transport at the blood-brain barrier. Brain Res 652(1): 76-82.
Roche‐Labarbe N, Zaaimi B, Berquin P, Nehlig A, Grebe R & Wallois F (2008) NIRS‐measured oxy‐and deoxyhemoglobin changes associated with EEG spike‐and‐wave discharges in children. Epilepsia 49(11): 1871-1880.
Rossi S, Telkemeyer S, Wartenburger I & Obrig H (2012) Shedding light on words and sentences: near-infrared spectroscopy in language research. Brain Lang 121(2): 152-163.
101
Rowley A, Payne S, Tachtsidis I, Ebden M, Whiteley J, Gavaghan D, Tarassenko L, Smith M, Elwell C & Delpy D (2006) Synchronization between arterial blood pressure and cerebral oxyhaemoglobin concentration investigated by wavelet cross-correlation. Physiol Meas 28(2): 161-173.
Rummel C, Zubler C, Schroth G, Gralla J, Hsieh K, Abela E, Hauf M, Meier N, Verma RK & Andres RH (2014) Monitoring cerebral oxygenation during balloon occlusion with multichannel NIRS. Journal of Cerebral Blood Flow & Metabolism 34(2): 347-356.
Sasai S, Homae F, Watanabe H, Sasaki AT, Tanabe HC, Sadato N & Taga G (2012) A NIRS–fMRI study of resting state network. Neuroimage 63: 179-193.
Sasai S, Homae F, Watanabe H & Taga G (2011) Frequency-specific functional connectivity in the brain during resting state revealed by NIRS. Neuroimage 56(1): 252-257.
Sassaroli A (2006) Spatially weighted BOLD signal for comparison of functional magnetic resonance imaging and near-infrared imaging of the brain. Neuroimage 33(2): 505-514.
Sato H, Kiguchi M, Kawaguchi F & Maki A (2004) Practicality of wavelength selection to improve signal-to-noise ratio in near-infrared spectroscopy. Neuroimage 21(4): 1554-1562.
Sato H, Yahata N, Funane T, Takizawa R, Katura T, Atsumori H, Nishimura Y, Kinoshita A, Kiguchi M, Koizumi H, Fukuda M & Kasai K (2013) A NIRS–fMRI investigation of prefrontal cortex activity during a working memory task. Neuroimage 83: 158-173.
Sawan M, Salam MT, Le Lan J, Kassab A, Gélinas S, Vannasing P, Lesage F, Lassonde M & Nguyen DK (2013) Wireless recording systems: From noninvasive EEG-NIRS to invasive EEG devices. Biomedical Circuits and Systems, IEEE Transactions on 7(2): 186-195.
Schelkanova I & Toronov V (2012) Independent component analysis of broadband near-infrared spectroscopy data acquired on adult human head. Biomedical optics express 3(1): 64-74.
Schelkanova I & Toronov V (2011) Principal and independent component analysis of concomitant functional near infrared spectroscopy and magnetic resonance imaging data. Proc. SPIE 8088, Diffuse Optical Imaging III. paper 80881M: 1-10.
Scholkmann F, Gerber U, Wolf M & Wolf U (2013) End-tidal CO 2: an important parameter for a correct interpretation in functional brain studies using speech tasks. Neuroimage 66: 71-79.
Scholkmann F, Kleiser S, Metz AJ, Zimmermann R, Pavia JM, Wolf U & Wolf M (2014) A review on continuous wave functional near-infrared spectroscopy and imaging instrumentation and methodology. Neuroimage 85: 6-27.
Schroeter ML, Kupka T, Mildner T, Uludağ K & von Cramon DY (2006) Investigating the post-stimulus undershoot of the BOLD signal—a simultaneous fMRI and fNIRS study. Neuroimage 30(2): 349-358.
Scofield JH (1994) Frequency-domain description of a lock-in amplifier. American Journal of Physics 62(2): 129-132.
102
Selb J, Boas DA, Chan S, Evans KC, Buckley EM & Carp SA (2014) Sensitivity of near-infrared spectroscopy and diffuse correlation spectroscopy to brain hemodynamics: simulations and experimental findings during hypercapnia. Neurophotonics 1(1): 015005-1-015005-15.
Shmueli K, van Gelderen P, de Zwart JA, Horovitz SG, Fukunaga M, Jansma JM & Duyn JH (2007) Low-frequency fluctuations in the cardiac rate as a source of variance in the resting-state fMRI BOLD signal. Neuroimage 38(2): 306-320.
Sirotin YB & Das A (2009) Anticipatory haemodynamic signals in sensory cortex not predicted by local neuronal activity. Nature 457(7228): 475-479.
Sitaram R, Zhang H, Guan C, Thulasidas M, Hoshi Y, Ishikawa A, Shimizu K & Birbaumer N (2007) Temporal classification of multichannel near-infrared spectroscopy signals of motor imagery for developing a brain–computer interface. Neuroimage 34(4): 1416-1427.
Slone E, Westwood E, Dhaliwal H, Federico P & Dunn JF (2012) Near-infrared spectroscopy shows preictal haemodynamic changes in temporal lobe epilepsy. Epileptic Disorders 14(4): 371-378.
Smith M (2011) Shedding light on the adult brain: a review of the clinical applications of near-infrared spectroscopy. Philos Trans A Math Phys Eng Sci 369(1955): 4452-4469.
Smith SM (2002) Fast robust automated brain extraction. Hum Brain Mapp 17(3): 143-155. Smith SM, Fox PT, Miller KL, Glahn DC, Fox PM, Mackay CE, Filippini N, Watkins KE,
Toro R, Laird AR & Beckmann CF (2009) Correspondence of the brain's functional architecture during activation and rest. Proc Natl Acad Sci U S A 106(31): 13040-13045.
Sokol D, Markand O, Daly E, Luerssen T & Malkoff M (2000) Near infrared spectroscopy (NIRS) distinguishes seizure types. Seizure 9(5): 323-327.
Sørensen E, Olesen J, Rask-Madsen J & Rask-andersen H (1978) The electrical potential difference and impedance between CSF and blood in unanesthetized man. Scandinavian Journal of Clinical & Laboratory Investigation 38(3): 203-207.
Sorvoja H, Myllylä T, Kirillin MY, Sergeeva EA, Myllylä RA, Elseoud A, Nikkinen J, Tervonen O & Kiviniemi V (2010) Non-invasive, MRI-compatible fibreoptic device for functional near-IR reflectometry of human brain. Quantum Electronics 40(12): 1067-1074.
Sorvoja H (2006) Noninvasive blood pressure pulse detection and blood pressure determination. PhD thesis. University of Oulu.
Steinbrink J, Villringer A, Kempf F, Haux D, Boden S & Obrig H (2006) Illuminating the BOLD signal: combined fMRI–fNIRS studies. Magn Reson Imaging 24(4): 495-505.
Steinhoff BJ, Herrendorf G & Kurth C (1996) Ictal near infrared spectroscopy in temporal lobe epilepsy: a pilot study. Seizure 5(2): 97-101.
Strangman G, Goldstein R, Rauch SL & Stein J (2006) Near-infrared spectroscopy and imaging for investigating stroke rehabilitation: test-retest reliability and review of the literature. Arch Phys Med Rehabil 87(12): 12-19.
Strangman G, Boas DA & Sutton JP (2002a) Non-invasive neuroimaging using near-infrared light. Biol Psychiatry 52(7): 679-693.
103
Strangman G, Culver JP, Thompson JH & Boas DA (2002b) A quantitative comparison of simultaneous BOLD fMRI and NIRS recordings during functional brain activation. Neuroimage 17(2): 719-731.
Strangman G, Franceschini MA & Boas DA (2003) Factors affecting the accuracy of near-infrared spectroscopy concentration calculations for focal changes in oxygenation parameters. Neuroimage 18(4): 865-879.
Suzuki M, Miyai I, Ono T & Kubota K (2008) Activities in the frontal cortex and gait performance are modulated by preparation. An fNIRS study. Neuroimage 39(2): 600-607.
Suzuki M, Miyai I, Ono T, Oda I, Konishi I, Kochiyama T & Kubota K (2004) Prefrontal and premotor cortices are involved in adapting walking and running speed on the treadmill: an optical imaging study. Neuroimage 23(3): 1020-1026.
Tak S, Jang J, Lee K & Ye JC (2010) Quantification of CMRO2 without hypercapnia using simultaneous near-infrared spectroscopy and fMRI measurements. Phys Med Biol 55(11): 3249.
Tak S, Yoon SJ, Jang J, Yoo K, Jeong Y & Ye JC (2011) Quantitative analysis of hemodynamic and metabolic changes in subcortical vascular dementia using simultaneous near-infrared spectroscopy and fMRI measurements. Neuroimage 55(1): 176-184.
Tallgren P, Vanhatalo S, Kaila K & Voipio J (2005) Evaluation of commercially available electrodes and gels for recording of slow EEG potentials. Clinical Neurophysiology 116(4): 799-806.
Thompson GJ, Pan WJ & Keilholz SD (2015) Different dynamic resting state fMRI patterns are linked to different frequencies of neural activity. J Neurophysiol 114(1): 114-124.
Thompson GJ, Pan W-, Magnuson ME, Jaeger D & Keilholz SD (2014) Quasi-periodic patterns (QPP): Large-scale dynamics in resting state fMRI that correlate with local infraslow electrical activity. Neuroimage 84: 1018-1031.
Tisdall MM, Tachtsidis I, Leung TS, Elwell CE & Smith M (2007) Near-infrared spectroscopic quantification of changes in the concentration of oxidized cytochrome c oxidase in the healthy human brain during hypoxemia. J Biomed Opt 12(2): 024002-1-024002-7.
Tomita Y, Vialatte F, Dreyfus G, Mitsukura Y, Bakardjian H & Cichocki A (2014) Bimodal BCI using simultaneously NIRS and EEG. Biomedical Engineering, IEEE Transactions on 61(4): 1274-1284.
Tong Y, Hocke LM, Fan X, Janes AC & deB Frederick B (2015) Can apparent resting state connectivity arise from systemic fluctuations? Frontiers in human neuroscience 9(285): 1-13.
Tong Y, Hocke LM, Nickerson LD, Licata SC, Lindsey KP & Frederick BB (2013) Evaluating the effects of systemic low frequency oscillations measured in the periphery on the independent component analysis results of resting state networks. Neuroimage 76: 202-215.
104
Tong Y, Bergethon PR & Frederick Bd (2011) An improved method for mapping cerebrovascular reserve using concurrent fMRI and near-infrared spectroscopy with Regressor Interpolation at Progressive Time Delays (RIPTiDe). Neuroimage 56(4): 2047-2057.
Tong Y & Frederick Bd (2010) Time lag dependent multimodal processing of concurrent fMRI and near-infrared spectroscopy (NIRS) data suggests a global circulatory origin for low-frequency oscillation signals in human brain. Neuroimage 53(2): 553-564.
Toronov V, Myllylä T, Kiviniemi V & Tuchin V (2013) Dynamics of the brain: Mathematical models and non-invasive experimental studies. The European Physical Journal Special Topics 222(10): 2607-2622.
Toronov V, Walker S, Gupta R, Choi JH, Gratton E, Hueber D & Webb A (2003) The roles of changes in deoxyhemoglobin concentration and regional cerebral blood volume in the fMRI BOLD signal. Neuroimage 19(4): 1521-1531.
Toronov VY, Zhang X & Webb AG (2007) A spatial and temporal comparison of hemodynamic signals measured using optical and functional magnetic resonance imaging during activation in the human primary visual cortex. Neuroimage 34(3): 1136-1148.
Toronov V, Webb A, Choi JH, Wolf M, Safonova L, Wolf U & Gratton E (2001a) Study of local cerebral hemodynamics by frequency-domain near-infrared spectroscopy and correlation with simultaneously acquired functional magnetic resonance imaging. Optics Express 9(8): 417-427.
Toronov V, Webb A, Choi JH, Wolf M, Michalos A, Gratton E & Hueber D (2001b) Investigation of human brain hemodynamics by simultaneous near-infrared spectroscopy and functional magnetic resonance imaging. Med Phys 28(4): 521-527.
Torricelli A, Contini D, Pifferi A, Caffini M, Re R, Zucchelli L & Spinelli L (2014) Time domain functional NIRS imaging for human brain mapping. Neuroimage 85, Part 1: 28-50.
Tuchin V (2007) Tissue optics: light scattering methods and instruments for medical diagnosis. Bellingham, Washington, USA, SPIE press.
Uludag K, Kohl M, Steinbrink J, Obrig H & Villringer A (2002) Cross talk in the Lambert-Beer calculation for near-infrared wavelengths estimated by Monte Carlo simulations. J Biomed Opt 7(1): 51-59.
Uludağ K, Steinbrink J, Villringer A & Obrig H (2004) Separability and cross talk: optimizing dual wavelength combinations for near-infrared spectroscopy of the adult head. Neuroimage 22(2): 583-589.
Uludag K, Steinbrink J, Kohl-Bareis M, Wenzel RU, Villringer A & Obrig H (2004) Cytochrome-c-oxidase redox changes during visual stimulation measured by near-infrared spectroscopy cannot be explained by a mere cross talk artefact. Neuroimage 22(1): 109-119.
Van Beek AH, Claassen JA, Rikkert MGO & Jansen RW (2008) Cerebral autoregulation: an overview of current concepts and methodology with special focus on the elderly. Journal of Cerebral Blood Flow & Metabolism 28(6): 1071-1085.
105
Van der Zee P, Cope M, Arridge S, Essenpreis M, Potter L, Edwards A, Wyatt J, McCormick D, Roth S & Reynolds E (1992) Experimentally measured optical pathlengths for the adult head, calf and forearm and the head of the newborn infant as a function of inter optode spacing. In: Goldstick TK, McCabe M & Maguire DJ (eds) Oxygen Transport to Tissue XIII. New York, Springer US: 143-153.
Van Dijk KR, Hedden T, Venkataraman A, Evans KC, Lazar SW & Buckner RL (2010) Intrinsic functional connectivity as a tool for human connectomics: theory, properties, and optimization. J Neurophysiol 103(1): 297-321.
Vanhatalo S, Palva JM, Holmes MD, Miller JW, Voipio J & Kaila K (2004) Infraslow oscillations modulate excitability and interictal epileptic activity in the human cortex during sleep. Proc Natl Acad Sci U S A 101(14): 5053-5057.
Vanhatalo S, Tallgren P, Becker C, Holmes MD, Miller JW, Kaila K & Voipio J (2003) Scalp-recorded slow EEG responses generated in response to hemodynamic changes in the human brain. Clinical Neurophysiology 114(9): 1744-1754.
Vasdekis SN, Tsivgoulis G, Athanasiadis D, Andrikopoulou A, Voumvourakis K, Lazaris AM & Stamboulis E (2012) Cerebrovascular reacivity assessment in patients with carotid artery disease: a combined TCD and NIRS study. Journal of Neuroimaging 22(3): 261-265.
Vermeij A, Meel-van den Abeelen, Aisha SS, Kessels RP, van Beek AH & Claassen JA (2014) Very-low-frequency oscillations of cerebral hemodynamics and blood pressure are affected by aging and cognitive load. Neuroimage 85: 608-615.
Vern BA, Leheta BJ, Juel VC, Laguardia J, Graupe P & Schuette WH (1997) Interhemispheric synchrony of slow oscillations of cortical blood volume and cytochrome aa3 redox state in unanesthetized rabbits. Brain Res 775(1-2): 233-239.
Vernieri F, Tibuzzi F, Pasqualetti P, Altamura C, Palazzo P, Rossini P & Silvestrini M (2008) Increased cerebral vasomotor reactivity in migraine with aura: an autoregulation disorder? A transcranial Doppler and near‐infrared spectroscopy study. Cephalalgia 28(7): 689-695.
Villringer A, Planck J, Hock C, Schleinkofer L & Dirnagl U (1993) Near infrared spectroscopy (NIRS): a new tool to study hemodynamic changes during activation of brain function in human adults. Neurosci Lett 154(1): 101-104.
Villringer A, Planck J, Stodieck S, Bötzel K, Schleinkofer L & Dirnagl U (1994) Noninvasive assessment of cerebral hemodynamics and tissue oxygenation during activation of brain cell function in human adults using near infrared spectroscopy. In: Vaupel P, Zander R & Bruley DF (eds) Oxygen Transport to Tissue XV. New York, Springer US: 559-565.
Villringer K, Minoshima S, Hock C, Obrig H, Ziegler S, Dirnagl U, Schwaiger M & Villringer A (1997) Assessment of local brain activation A simultaneous PET and Near-Infrared Spectroscopy Study. In: Villringer A & Dirnagl U (eds) Optical Imaging of Brain Function and Metabolism 2. New York, Springer US: 149-153.
Virtanen J, Noponen T & Meriläinen P (2009) Comparison of principal and independent component analysis in removing extracerebral interference from near-infrared spectroscopy signals. J Biomed Opt 14(5): 054032-1-054032-10.
106
Virtanen J, Noponen T, Salmi T, Toppila J & Meriläinen P (2012) Impaired cerebral vasoreactivity may cause cerebral blood volume dip following obstructive sleep apnea termination. Sleep and Breathing 16(2): 309-312.
Voipio J, Tallgren P, Heinonen E, Vanhatalo S & Kaila K (2003) Millivolt-scale DC shifts in the human scalp EEG: Evidence for a nonneuronal generator. J Neurophysiol 89(4): 2208-2214.
Wallois F, Mahmoudzadeh M, Patil A & Grebe R (2012) Usefulness of simultaneous EEG–NIRS recording in language studies. Brain Lang 121(2): 110-123.
Watanabe A, Matsuo K, Kato N & Kato T (2003) Cerebrovascular response to cognitive tasks and hyperventilation measured by multi-channel near-infrared spectroscopy. J Neuropsychiatry Clin Neurosci 15(4): 442-449.
Watanabe E, Maki A, Kawaguchi F, Yamashita Y, Koizumi H & Mayanagi Y (1998) Noninvasive cerebral blood volume measurement during seizures using multichannel near infrared spectroscopic topography. Journal of Epilepsy 11(6): 335-340.
Watanabe E, Nagahori Y & Mayanagi Y (2002) Focus diagnosis of epilepsy using near‐infrared spectroscopy. Epilepsia 43(s9): 50-55.
White BR, Snyder AZ, Cohen AL, Petersen SE, Raichle ME, Schlaggar BL & Culver JP (2009) Resting-state functional connectivity in the human brain revealed with diffuse optical tomography. Neuroimage 47(1): 148-156.
Wise RG, Ide K, Poulin MJ & Tracey I (2004) Resting fluctuations in arterial carbon dioxide induce significant low frequency variations in BOLD signal. Neuroimage 21(4): 1652-1664.
Wolf M, Ferrari M & Quaresima V (2007) Progress of near-infrared spectroscopy and topography for brain and muscle clinical applications. J Biomed Opt 12(6): 062104-1-062104-14.
Wolf M, Wolf U, Toronov V, Michalos A, Paunescu LA, Choi JH & Gratton E (2002) Different time evolution of oxyhemoglobin and deoxyhemoglobin concentration changes in the visual and motor cortices during functional stimulation: a near-infrared spectroscopy study. Neuroimage 16(3): 704-712.
Woody CD, Marshall WH, Besson JM, Thompson HK, Aleonard P & Albe-Fessard D (1970) Brain potential shift with respiratory acidosis in the cat and monkey. Am J Physiol 218(1): 275-283.
Wróbel M, Gnyba M, Jedrzejewska-Szczerska M, Myllyla T, Smulko J & Barman I (2015a) Sensing of anesthetic drugs in blood with Raman spectroscopy. Proc. OSA Technical Digest, Advanced Photonics 2015. paper SeS1B.4: 1-3.
Wróbel M, Gnyba M, Urniaż R, Myllylä T & Jędrzejewska-Szczerska M (2015b) Detection of propofol concentrations in blood by Raman spectroscopy. Proc. SPIE9537, Clinical and Biomedical Spectroscopy and Imaging IV. paper 95370Z: 1-7.
Wyatt J, Cope M, Delpy D, Van der Zee P, Arridge S, Edwards A & Reynolds E (1990) Measurement of optical path length for cerebral near-infrared spectroscopy in newborn infants. Dev Neurosci 12(2): 140-144.
107
Wylie GR, Graber HL, Voelbel GT, Kohl AD, DeLuca J, Pei Y, Xu Y & Barbour RL (2009) Using co-variations in the Hb signal to detect visual activation: a near infrared spectroscopic imaging study. Neuroimage 47(2): 473-481.
Xu Y, Graber HL, Barbour RL, Blanco I, Zirak P, Fortuna A, Cotta G, Mayos M, Mola A & Durduran T (2014) nirsLAB: A Computing Environment for fNIRS Neuroimaging Data Analysis. Proc. OSA Technical Digest, Biomedical Optics 2014. paper BM3A.1.
Yamashita Y, Maki A & Koizumi H (2001) Wavelength dependence of the precision of noninvasive optical measurement of oxy-, deoxy-, and total-hemoglobin concentration. Med Phys 28(6): 1108-1114.
Yang R, Brugniaux J, Dhaliwal H, Beaudin AE, Eliasziw M, Poulin MJ & Dunn JF (2015) Studying cerebral hemodynamics and metabolism using simultaneous near-infrared spectroscopy and transcranial Doppler ultrasound: a hyperventilation and caffeine study. Physiol Rep 3(4): e12378, 1-11.
Ye JC, Tak S, Jang KE, Jung J & Jang J (2009) NIRS-SPM: statistical parametric mapping for near-infrared spectroscopy. Neuroimage 44(2): 428-447.
Yoshino K, Oka N, Yamamoto K, Takahashi H & Kato T (2013a) Correlation of prefrontal cortical activation with changing vehicle speeds in actual driving: a vector-based functional near-infrared spectroscopy study. Front Hum Neurosci 7(895): 1-9.
Yoshino K, Oka N, Yamamoto K, Takahashi H & Kato T (2013b) Functional brain imaging using near-infrared spectroscopy during actual driving on an expressway. Front Hum Neurosci 7(882): 1-16.
Yücel MA, Evans KC, Selb J, Huppert TJ, Boas DA & Gagnon L (2014) Validation of the hypercapnic calibrated fMRI method using DOT–fMRI fusion imaging. Neuroimage 102: 729-735.
Zahneisen B, Hugger T, Lee KJ, Levan P, Reisert M, Lee H-, Assländer J, Zaitsev M & Hennig J (2012) Single shot concentric shells trajectories for ultra fast fMRI. Magn Reson Med 68(2): 484-494.
Zaidi AD, Munk MH, Schmidt A, Risueno-Segovia C, Bernard R, Fetz E, Logothetis N, Birbaumer N & Sitaram R (2015) Simultaneous epidural functional near-infrared spectroscopy and cortical electrophysiology as a tool for studying local neurovascular coupling in primates. Neuroimage 120: 394-399.
Zhang Y, Lu C, Biswal BB, Zang Y, Peng D & Zhu C (2010) Detecting resting-state functional connectivity in the language system using functional near-infrared spectroscopy. J Biomed Opt 15(4): 047003-1-047003-8.
Zhao H, Tanikawa Y, Gao F, Onodera Y, Sassaroli A, Tanaka K & Yamada Y (2002) Maps of optical differential pathlength factor of human adult forehead, somatosensory motor and occipital regions at multi-wavelengths in NIR. Phys Med Biol 47(12): 2075-2093.
Zheng F, Sheinberg R, Yee MS, Ono M, Zheng Y & Hogue CW (2013) Cerebral near-infrared spectroscopy monitoring and neurologic outcomes in adult cardiac surgery patients: a systematic review. Anesth Analg 116(3): 663-676.
108
Zhou Y, Lui YW, Zuo X, Milham MP, Reaume J, Grossman RI & Ge Y (2014) Characterization of thalamo‐cortical association using amplitude and connectivity of functional MRI in mild traumatic brain injury. Journal of Magnetic Resonance Imaging 39(6): 1558-1568.
Zourabian A, Siegel A, Chance B, Ramanujam N, Rode M & Boas DA (2000) Trans-abdominal monitoring of fetal arterial blood oxygenation using pulse oximetry. J Biomed Opt 5(4): 391-405.
109
Original publications
I Korhonen V, Myllylä T, Kirillin M, Popov A, Bykov A, Gorshkov A, Sergeeva A, Kinnunen M & Kiviniemi V (2014) Light Propagation in NIR Spectroscopy of the Human Brain. IEEE Journal of Selected Topics in Quantum Electronics 20(2): 7100310.
II Myllylä T, Korhonen V, Surazynski L, Zienkiewicz A, Sorvoja H & Myllylä R (2014) Measurement of cerebral blood flow and metabolism using light-emitting diodes. Measurement 58(December 2014): 387-393.
III Korhonen V, Hiltunen T, Myllylä T, Wang X, Kantola J, Nikkinen J, Zang Y-F, Levan P & Kiviniemi V (2014) Synchronous multi-scale neuroimaging environment for critically sampled physiological analysis of brain function – Hepta-scan concept. Brain Connectivity 4(9): 677–689.
IV Kiviniemi V, Korhonen V, Kortelainen J, Rytky S, Keinänen (nee Hiltunen) T, Tuovinen T, Isokangas M, Sonkajärvi E, Siniluoto T, Nikkinen J, Alahuhta S, Tervonen O, Turpeenniemi-Hujanen T, Myllylä T, Kuittinen O & Voipio J (2015) Real-time monitoring of human blood brain barrier disruption. Manuscript.
Reprinted with permission from IEEE (I), Elsevier (II) and Mary Ann Liebert, Inc.
(III).
Original publications are not included in the electronic version of the dissertation.
110
A C T A U N I V E R S I T A T I S O U L U E N S I S
Book orders:Granum: Virtual book storehttp://granum.uta.fi/granum/
S E R I E S D M E D I C A
1377. Väyrynen, Sara (2016) Histological and molecular features of serrated colorectaladenocarcinoma and its precursor lesions
1378. Kujanpää, Kirsi (2016) Mechanisms behind stem cell therapy in acute myocardialinfarction
1379. Törmälä, Reeta-Maria (2016) Human zona pellucida abnormalities : a geneticapproach to the understanding of fertilization failure
1380. Pinola, Pekka (2016) Hyperandrogenism, menstrual irregularities and polycysticovary syndrome : impact on female reproductive and metabolic health from earlyadulthood until menopause
1381. Haghighi Poodeh, Saeid (2016) Novel pathomechanisms of intrauterine growthrestriction in fetal alcohol syndrome in a mouse model
1382. Helminen, Olli (2016) Glucose metabolism in preclinical type 1 diabetes
1383. Raatiniemi, Lasse (2016) Major trauma in Northern Finland
1384. Huhta, Heikki (2016) Toll-like receptors in Alimentary tract -special reference toBarrett’s esophagus
1385. Kelhälä, Hanna-Leena (2016) The effect of systemic treatment on immuneresponses and skin microbiota in acne
1386. Lappalainen, Olli-Pekka (2016) Healing of cranial critical sized defects with grafts,stem cells, growth factors and bio-materials
1387. Ronkainen, Justiina (2016) Role of Fto in the gene and microRNA expression ofmouse adipose tissues in response to high-fat diet
1388. Ronkainen, Eveliina (2016) Early risk factors influencing lung function inschoolchildren born preterm in the era of new bronchopulmonary dysplasia
1389. Casula, Victor (2016) Quantitative magnetic resonance imaging methods forevaluation of articular cartilage in knee osteoarthritis : free-precession androtating-frame relaxation studies at 3 Tesla
1390. Alahuhta, Ilkka (2016) The microenvironment is essential for OTSCCprogression
1391. Hagnäs, Maria (2016) Health behavior of young adult men and the associationwith body composition and physical fitness during military service
1392. Korhonen, Vesa (2016) Integrating near-infrared spectroscopy to synchronousmultimodal neuroimaging : Applications and novel findings
UNIVERSITY OF OULU P .O. Box 8000 F I -90014 UNIVERSITY OF OULU FINLAND
A C T A U N I V E R S I T A T I S O U L U E N S I S
Professor Esa Hohtola
University Lecturer Santeri Palviainen
Postdoctoral research fellow Sanna Taskila
Professor Olli Vuolteenaho
University Lecturer Veli-Matti Ulvinen
Director Sinikka Eskelinen
Professor Jari Juga
University Lecturer Anu Soikkeli
Professor Olli Vuolteenaho
Publications Editor Kirsti Nurkkala
ISBN 978-952-62-1413-9 (Paperback)ISBN 978-952-62-1414-6 (PDF)ISSN 0355-3221 (Print)ISSN 1796-2234 (Online)
U N I V E R S I TAT I S O U L U E N S I S
MEDICA
ACTAD
D 1392
ACTA
Vesa Korhonen
OULU 2016
D 1392
Vesa Korhonen
INTEGRATING NEAR-INFRARED SPECTROSCOPY TO SYNCHRONOUS MULTIMODAL NEUROIMAGINGAPPLICATIONS AND NOVEL FINDINGS
UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF MEDICINE;MEDICAL RESEARCH CENTER OULU;OULU UNIVERSITY HOSPITAL