myocardial perfusion and function

ASNC IMAGING GUIDELINES FOR NUCLEAR
CARDIOLOGY PROCEDURES
Myocardial perfusion and function: Single photonemission computed tomography

Christopher L. Hansen, MD,a Richard A. Goldstein, MD,a

Olakunle O. Akinboboye, MBBS, MPH, MBA,b Daniel S. Berman, MD,b

Elias H. Botvinick, MD,b Keith B. Churchwell, MD,b C. David Cooke, MSEE,b

James R. Corbett, MD,b S. James Cullom, PhD,b Seth T. Dahlberg, MD,b

Regina S. Druz, MD,b Edward P. Ficaro, PhD,b James R. Galt, PhD,b Ravi K. Garg, MD,b

Guido Germano, PhD,b Gary V. Heller, MD, PhD,b Milena J. Henzlova, MD,b

Mark C. Hyun, CNMT, NCT, RT(N)(R),b Lynne L. Johnson, MD,b

April Mann, CNMT, NCT, RT(N),b Benjamin D. McCallister, Jr, MD,b

Robert A. Quaife, MD,b Terrence D. Ruddy, MD,b Senthil N. Sundaram, MD, MPH,b

Raymond Taillefer, MD,b R. Parker Ward, MD,b John J. Mahmarian, MDc

Purpose. To evaluate regional myocardial perfusionand function.

ACQUISITION PROTOCOLS

Protocols for the various nuclear cardiology singlephoton emission computed tomography (SPECT) acqui-sition studies are presented in the following pages(Tables 1-6). For each of the protocols, the acquisitionparameters are listed along with their correspondingvalue for exercise and rest. Implementation of theseprotocol acquisition parameters has been shown to pro-vide acceptable images of good quality for routineclinical interpretation and quantitation. However, proto-col parameters other than those listed may be preferred atsome institutions, and ongoing research into correctionsfor attenuation, scatter, and camera response depth de-pendence may result in better parameters in the future.Therefore these protocols should be viewed as a consen-sus of opinion on the parameters that will provideacceptable images. A description for each of the acqui-sition parameters is listed below.

1. Dose. The doses for each of the protocols representstandard doses commonly used clinically. The stan-dard doses described are given for an average 70-kgpatient. Doses may be adjusted upward for heavierpatients by 0.04 mCi/kg for thallium 201 and by 0.31mCi/kg for technetium 99m. Other options are in-

Co-Chaira Membera Board Reviewera

J Nucl Cardiol 2007;14:e39-60.1071-3581/$32.00Copyright © 2007 by the American Society of Nuclear Cardiology.
doi:10.1016/j.nuclcard.2007.09.023
creased imaging times, the use of multidetectorsystems, or 2-day imaging. Tl-201 imaging timescan be adjusted based on the counts acquired for apreliminary 4-minute planar study in order to ensureacquiring at least 500,000 background-subtractedmyocardial counts.

2. Position. Factors influencing patient position in-clude camera/gantry design, minimization of arti-facts, and patient comfort. The supine position isroutinely used for SPECT imaging with most cur-rently available systems and protocols. Prone imag-ing has been reported to produce less patient motionand less inferior wall attenuation than supine imag-ing.1,2 The combination of supine and prone imagesmay be helpful in identifying attenuation artifactsdue to breast and/or excessive lateral chest-wall fat,due to the shift in position of the attenuating struc-tures that occurs in the prone position. In somelaboratories the advantages of prone imaging inclarifying artifactual defects has led to a routine useof the combination of supine followed by proneacquisitions.3 It appears that prone imaging does noteliminate attenuation artifact but rather simplychanges the location. By comparing supine andprone images, artifactual defects will change theirlocation whereas true perfusion defects will remainfixed.4,5 Therefore it is suggested that prone imagingbe done in combination with supine imaging and notsimply replace it. When being used in this fashion,the acquisition time for the secondary (prone) imageset is reduced by 20% to 40%. Using a dual-detectorcamera with 25 to 30 mCi of a Tc-99m perfusionagent, supine acquisitions are performed for 25
seconds per stop and prone acquisitions for 15
e39

e40 Hansen et al Journal of Nuclear CardiologyMyocardial perfusion and function: SPECT November/December 2007

seconds per stop, with 30 to 32 stops per detectorbeing obtained per acquisition along a 180° orbit(right anterior oblique 45° to left posterior oblique45°). With Tl-201, imaging should be begun approx-imately 10 to 15 minutes after stress testing, and ifsoft-tissue attenuation or patient motion compro-mises a study, the benefit of repeating the acquisitionis questionable. In contrast, Tc-99m sestamibi orTc-99m tetrofosmin permits stress testing and tracerinjection to take place at a location remote from theimaging laboratory and image acquisition can sim-ply be repeated when patient motion, soft-tissueattenuation, or other artifact is considered to beresponsible for the production of a perfusion defect.Some camera/gantry designs require the patient to bepositioned in a more upright position. Changes inpatient positioning from those described above willlikely cause changes in the distribution of adjacentsoft-tissue attenuation and need to be considered inimage interpretation. New normal databases willmost likely need to be generated for different patient

Table 1. Patient protocol: Same-day rest-stress Tc-99

Rest study

Dose 8-12 mCiPosition Supine

ProneUpright/semi-upright

Delay time (intervals)Injection ¡ imaging 30-60 minRest ¡ stress

Acquisition protocolEnergy window 15%-20% symmetricCollimator LEHROrbit 180° (45° RAO to 45° LPO)Orbit type Circular

Non-circularPixel size 6.4 � 0.4 mmAcquisition type Step and shoot

ContinuousNo. of projections 60-64Matrix 64 � 64Time/projection 25 sECG gated OptionalFrames/cycle 8

816

R-to-R window 100%

RAO, Right anterior oblique; LPO, left posterior oblique.

positions.

3. Delay time. These times are listed as ranges; spe-cific values are optional. The objectives are to allowthe patient to recover fully from exercise, thusallowing heart rate to return to baseline (reducinggating artifact), avoiding “upward creep” fromchanges in respiratory patterns while dyspnea re-solves, and to minimize interference from hepaticuptake.6 Provided that imaging times fall within thespecified ranges, clinically useful SPECT imagesshould result.

4. Energy windows. Energy window position is deter-mined by the radioisotope employed, 140 keV fortechnetium-based perfusion agents and 70 keV forthallium. It is reasonable to simultaneously acquirethe higher energy peaks of thallium (135 and 167keV) on cameras that are capable of doing this. Thewindow sizes are determined largely by conventionand reflect the tradeoff between image counts andresolution. The values shown are the most com-monly used and have been found to be acceptable formost cameras. On systems offering improved energy

quisition

tress studyFor information,see paragraph

36 mCi Standard 1ine Standard 2ne Optional 2ight/semi-upright Optional

60 min Standard 3min to 4 h Standard 3

e Standard 4e Preferred 5e Preferred 6e Standard 7e Standard 7e Standard 8e Standard 9e Optional 9e Standard 10e Standard 11

s Standard 12ndard Preferred 14

Standard 14OptionalOptional 14

% Preferred 14

m ac

S

24-SupProUpr

15-30

SamSamSamSamSamSamSamSamSamSam20Sta81616100

resolution, the window size may be reduced, result-

Journal of Nuclear Cardiology Hansen et al e41Volume 14, Number 6;e39-60 Myocardial perfusion and function: SPECT

ing in decreased scatter and improved image reso-lution, so long as imaging times are extended toacquire the same clinically useful number of counts.The same energy windows used in performing pa-tient studies should be used for routine daily qualitycontrol (QC).

5. Collimator. Parallel-hole collimators are most com-monly employed for cardiac SPECT acquisitions.They fall into two categories: low-energy all-pur-pose (LEAP), used mostly for Tl-201 studies, andlow-energy high-resolution (LEHR), used for Tc-99m studies. Compared with LEAP collimators,LEHR collimators have longer bores, thinner septa,and smaller holes, which provide better resolution atthe expense of reduced sensitivity. Therefore, to useLEHR collimators, imaging agents providing highcount rates are required (ie, Tc-99m agents). Gener-ally, LEAP collimators are used for 3-mCi Tl-201studies, including gated SPECT acquisitions. Fordual-isotope studies, LEHR collimators are sug-gested.

6. Orbit. Due to the anterior position of the heart in the

Table 2. Patient protocol: Same-day stress-rest Tc-99

Stress study

Dose 8-12 mCiPosition Supine


Delay time (intervals)Injection ¡ imaging 15 min to 1 hStress ¡ rest

Acquisition protocolEnergy window 15%-20% symmetricCollimator LEHROrbit 180° (45° RAO to 45° LPO)Orbit type Circular


ContinuousNo. of projections 60-64Matrix 64 � 64Time/projection 25 sECG gated OptionalFrames/cycle 8

16R-to-R window 100%


left hemithorax, much higher count rates are ob-

tained per given period of imaging time for a 180°orbit (45° right anterior oblique to 45° left posterioroblique) compared to a 360° orbit.7 The recommen-dation of which orbit will depend on the cameraconfiguration; it does not seem to be worthwhile toincrease imaging time to complete a 360° orbit sincemuch better count statistics will be obtained if thattime is used to increase acquisition on a 180° orbit.8

A 360° orbit is appropriate for 3-headed cameraswith a 360° configuration where a 360° orbit isacquired in the same time as a 180° orbit. The utilityof the posterior 180° of a 360° orbit is much greaterfor higher-energy radioisotopes (such as technetium)compared to low-energy radioisotopes (such as thal-lium).

7. Orbit type. The main orbit options in SPECTmyocardial perfusion imaging are circular versusnoncircular (elliptical or body-contoured) orbits.Noncircular orbits follow the contour of the patient,bringing the camera closer to the patient, therebyimproving spatial resolution. Circular orbits main-tain a fixed radius of rotation and on average result

quisition

Rest studyFor information,see paragraph

36 mCi Standard 1ine Standard 2ne Optional 2ight/semi-upright Optional

60 min Standard 3min to 4 h Standard 3

e Standard 4e Preferred 5e Preferred 6e Standard 7e Standard 7e Standard 8e Standard 9e Optional 9e Standard 10e Standard 11


Standard 14Optional 14

% Preferred 14

m ac

24-SupProUpr

30-30

SamSamSamSamSamSamSamSamSamSam20Sta816100

in the detector being further from the patient. In


general, there is reduced (but more uniform) spatialresolution with circular orbits since the detector-to-source distance is greater with this technique. Cir-cular acquisitions continue to be the most frequentlyused option, but some manufacturers do providenoncircular orbit capability. Imaging artifacts havebeen observed when noncircular orbits are used, dueto increased variation of source-to-detector distance,resulting in variation of spatial resolution.9

8. Pixel size. The SPECT protocols listed here specifya 6.4 � 0.4–mm pixel size for a 64 � 64 imagematrix. This size offers satisfactory image resolutionfor interpretation and quantitation of both Tl-201and Tc-99m tomograms.

9. Acquisition type. The most widespread mode oftomographic acquisition is the “step-and-shoot”method. In this approach, the camera acquires aprojection and then stops recording data when mov-ing to the next angle; this results in a small amountof dead time since the camera is not acquiring datawhile it is moving. An alternative is “continuous”mode, where the camera moves continuously andacquires each projection over an angular increment.This eliminates dead time and thus increases image

Table 3. Patient protocol: Two-day stress Tc-99m acq

Stress study

Dose 30 mCiPosition Supine

ProneDelay time (intervals)

Injection ¡ imaging 15-60 minAcquisition protocol

Energy window 15%-20% symmetricCollimator LEHROrbit 180° (45° RAO to 45° LPO)Orbit type Circular


ContinuousNo. of projections 60-64Matrix 64 � 64Time/projection 20 sECG gated StandardFrames/cycle 8



counts at the expense of a small amount of blurring

due to the motion of the camera head while acquir-ing. It seems likely that the increase in countstatistics more than offsets the small amount ofblurring due to camera motion.

10. Number of projections. The optimal number ofprojections for emission studies depends on match-ing the number of projections to the resolution of thesystem. A thallium SPECT acquisition with a LEAPcollimator is a relatively low-resolution study, forwhich 32 projections over 180° are sufficient. Ahigher-resolution study using Tc-99m agents shouldbe collected with a high-resolution collimator; thisrequires at least 60 to 64 projections over 180° toprevent loss of resolution. Larger numbers of pro-jections are not necessary at this time but couldbecome beneficial if technical innovations result inimproved overall system resolution.

11. Matrix. The standard matrix size for emissionSPECT is 64 � 64 pixels.

12. Time/projection. The emission acquisition timeslisted have been found to produce images of accept-able and comparable quality for rest and stressstudies.

13. Total time. For single-detector systems, the total

on

Rest studyFor information,see paragraph

30 mCi Standard 1Supine Standard 2Prone Optional 2

30-60 min Standard 3

Same Standard 4Same Preferred 5Same Preferred 6Same Standard 7Same Standard 7Same Standard 8Same Standard 9Same Optional 9Same Standard 10Same Standard 1120 s Standard 12Standard Preferred 148 Standard 1416 Optional 14100% Preferred 14

uisiti

time for an emission acquisition ultimately is based


on how long a patient can tolerate the procedurewithout moving, balanced by the need to acquiresufficient counts. The maximum practical time is onthe order of 30 minutes. For 90° dual-detectorsystems, this time can be halved, and many labora-tories obtain gated perfusion SPECT studies in only12 to 15 minutes using biplane cameras. Consider-ation may be made for increasing imaging time inpatients likely to have lower count statistics (eg,obese patients) if it is felt that they can tolerate it.

14. Gated SPECT. Incorporation of wall motion andwall thickening information from gated SPECT hasbeen shown to increase specificity and confidence byhelping to differentiate breast and diaphragmaticattenuation artifacts from true perfusion defects.Likewise, assessment of regional wall motion and/orthickening can be a valuable tool for detectingviability within a stress-induced perfusion defect.Left ventricular (LV) ejection fractions (EFs) andvolumes, as well as regional wall motion and thick-ening, now are computed routinely from gated

Table 4. Patient protocol: Separate dual-isotope acqu

Rest study

Dose 2.5-3.5 mCi Tl-201Position Supine


Delay time (intervals)Injection ¡ imaging 10-15 minRest ¡ stress

Acquisition protocolEnergy window 25%-30% symmetric, 70 keV

20% symmetric, 167 keVCollimator LEHROrbit 180° (45° RAO to 45° LPO)Orbit type Circular


ContinuousNo. of projections 32-64Matrix 64 � 64Time/projection 40 s (32 fr), 25 s (64 fr)ECG gated OptionalFrames/cycle 8



SPECT data using commercially available soft-

ware.10 The majority of stress myocardial perfusionradionuclide studies currently are acquired as gatedSPECT data. However, there is mounting evidencethat the information content of the post-stress acqui-sition may be different from that of the resting datain patients who have post-ischemic stunning ofmyocardium.11 Discrepancies may also be present ifthe count density is suboptimal due to the lowertracer dose in the resting scan (same-day rest/stressprotocol) and/or hypoperfusion of a segment atstress. Providing that there is adequate count density,particularly with regard to the lower-dose acquisi-tions, both stress and rest SPECT perfusion studiesmay be acquired as gated data sets. Optimizingprotocols for which both stress and rest gated dataare acquired remains an area of investigation.

15. Multidetector systems. It is recommended for mul-tidetector systems that total imaging time be ad-justed to obtain greater than the minimum countslisted in the “Instrumentation Quality Assurance andPerformance” section of these guidelines (in the

n

Stress studyFor information,see paragraph

mCi Tc-99m Standard 1ine Standard 2ne Optionalright/semi-upright Optional

-60 min Standard 3delay Standard 3

%-20% symmetric,40 keV

Standard 4

e Preferred 5e Preferred 6e Standard 7e Standard 7e Standard 8e Standard 9e Optional 9

-64 Standard 10e Standard 11


Standard 14Optional

0% Preferred 14

isitio

30SupProUp

15No

151

SamSamSamSamSamSamSam60Sam20Sta81610

subsection entitled “Clinical QA for Each Patient

evalua


Procedure”) but less than a maximum total imagingtime of 30 minutes.

16. The following acquisition parameters are recom-mended for the imaging protocols described in the“Stress Protocols and Tracers” section of theseguidelines.

PROCESSING PROTOCOLS

1. Filtering. Image filtering is a very complex topic thatencompasses techniques for image enhancement, re-construction, and feature extraction.12,13 The mainarea of concern for an interpreter of SPECT studies isimage enhancement by reducing noise prior to imagereconstruction. All forms of imaging are plagued bystatistical variation in the acquired image countscommonly referred to as noise. The quality of animage can be described as the signal-to-noise ratio,which describes the relative strength of the signalcomponent (what is actually being imaged) comparedto noise. The signal-to-noise ratio is much higher atlower spatial frequencies (broad features that areconstant over many pixels) and decreases at higherspatial frequencies (features that change over few

Table 5. Patient protocol: Stress/redistribution Tl-201

Stress study

Dose 2.5-3.5 mCi Tl-201Position Supine


Delay time (intervals)Injection ¡ imaging 10-15 min*Stress ¡ rest

Acquisition protocolEnergy window 30% symmetric, 70 keV

20% symmetric, 167 keVCollimator LEAPOrbit 180° (45° RAO to 45° LPO)Orbit type Circular


ContinuousNo. of projections 32-64Matrix 64 � 64Time/projection 40 s (32 fr), 25 s (64 fr)

RAO, Right anterior oblique; LPO, left posterior oblique.*An anterior planar image may be acquired during this interval to

pixels such as edges). In general, the greater the count

statistics, the better the signal-to-noise ratio. A low-pass filter is generally used to reduce noise because itallows low spatial frequencies to pass through andattenuates the high frequencies where image noisepredominates. Low-pass filters such as the Hanningand Butterworth can be characterized by a cutofffrequency where they begin to affect the image. Thecutoff frequency can be adjusted, depending on thesignal-to-noise ratio, to preserve as much of the signaland suppress14 as much noise as possible. If the cutoffis too high, there is significant noise in the image; ifthe cutoff is too low, significant information in thesignal is suppressed. Nuclear cardiology images, be-cause of their relatively low count statistics, tend tohave greater amounts of image noise, and filteredbackprojection, because of its dependence on rampfiltering, tends to amplify this noise. The optimal filterfor a given image depends on the signal-to-noise ratiofor that image; under-filtering an image leaves signif-icant noise in the image, and over-filtering unneces-sarily blurs image detail; both over-filtering andunder-filtering can reduce image accuracy. Softwarereconstruction packages are set with default filterselection and cutoff values that are optimized for the

isition

istribution reststudy

For information,see paragraph

applicable Standard 1ne Standard 2e Optionalht/semi-upright Optional

applicable Standard 3Standard 3

e Standard 4

e Preferred 5e Preferred 6e Standard 7e Standard 7e Standard 8e Standard 9e Optional 9e Standard 10e Standard 11(32 fr), 25 s (64 fr) Standard 12

te Tl-201 lung uptake.

acqu

Red

NotSupiPronUprig

Not3-4 h

Sam

SamSamSamSamSamSamSamSamSam40 s

average patient. Adjustment of the filter cutoff can be


done in patients with poor count statistics (eg, obesepatients) to optimally filter their images. However,this is discouraged unless the physician is thoroughlyfamiliar with filter adjustment and the potential ef-fects. Changing the filter cutoff may have unexpectedeffects on the output of commercially available anal-ysis programs, especially those that employ edgedetection such as defect quantitation and LV volumesand EF. Deconvolving filters, such as the Metz andWiener filters, can correct for blurring that occursfrom scatter as photons travel through the body.Although images may look sharper with these filters,these filters have not yet been shown to improveimage accuracy.12

2. Reconstruction. The traditional method of imagereconstruction has been filtered backprojection, atechnique based on a mathematical proof, whichassumes no attenuation, no scatter, and an infinitenumber of projections. It is relatively straightforward

Table 6. Patient protocol: Stress/reinjection/redistribu

Stress study Reinjec

Dose 2.5-3.5 mCi 1.0-1.5mPosition Supine


Delay time (intervals)Injection ¡ imaging 10-15 minStress ¡ redistributionReinjection ¡ imaging

(MI)24-h imaging

Acquisition protocolEnergy window 30% symmetric,

70 keV20% symmetric,

167 keVCollimator LEAPOrbit 180° (45° RAO to

45° LPO)Orbit type Circular


ContinuousNo. of projections 32-64Matrix 64 � 64Time/projection 40 s (32 fr), 25 s

(64 fr)


and comparatively fast.15 The vast majority of clinical

experience is based upon it, and it has withstood thetest of time despite its inability to model attenuationand scatter. There is a different class of reconstructionalgorithms that are based on iterative techniques.These algorithms start with a rudimentary guess of thedistribution, generate projections from the guess, andcompare these projections to the acquired projections.The guess is refined based on the differences betweenthe generated and actual projections, and the processis repeated (hence the term “iterative”) usually for afixed number of iterations but can also be repeateduntil the error between the generated and actualprojections is acceptably small. A main advantage ofthese algorithms is that the process of generatingprojections from the guess can be made as sophisti-cated as desired and can incorporate such variables asattenuation, scatter, and depth-dependent blur. Themain disadvantage is the computational intensity ofthe algorithm; it takes many times longer to complete

Tl-201 acquisition

(Redistribution)rest study


Not applicable Standard 1Supine Standard 2Prone Optional 2Upright/semi-upright Optional

Not applicable Standard 33-4 h Standard 320-30 min Standard 3

Optional 3

Same Standard 4

Same Preferred 5Same Preferred 6

Same Standard 7Same Standard 7Same Standard 8Same Standard 9Same Optional 9Same Standard 10Same Standard 1140 s (32 fr), 25 s

(64 fr)Standard 12

tion

tion

Ci

than filtered backprojection. However, due to contin-


ual increases in computer processor speed, thesealgorithms can now be completed in an acceptabletime for routine clinical use. Nonetheless, iterativetechniques have not yet been proven to be unequivo-cally superior to filtered backprojection.

3. Reorientation. A critical phase of myocardial pro-cessing is reorientation of tomographic data into thenatural approximate symmetry axes of an individualpatient’s heart. This is performed either by an ob-server or automatically and results in sectioning thedata into vertical long-axis, horizontal long-axis, andshort-axis planes. Long-axis orientation lines shouldbe parallel to long-axis walls of the myocardium andshould be consistent between rest and stress studies.Inappropriate plane selections can result in mis-aligned myocardial walls between rest and stress datasets, potentially resulting in incorrect interpretation. Itis crucial that all axis choices be available as QCscreens, and that these are reviewed by the technolo-gist and the physician who reads each study to verifythat axes were selected properly.

4. Display–cine review. The most important post-ac-quisition QC procedure is to view the raw tomo-graphic data in cine mode. This presentation offers asensitive method for detecting patient and/or heartmotion, “upward creep,” breast shadow due to atten-uation, diaphragmatic attenuation, and superimposedabdominal visceral activity, all of which can createartifacts in the reconstructed images. Review of theraw tomograms in cine mode is performed twice:once by the technologist immediately after the acqui-sition and again by the physician during imageinterpretation. For gated studies, usually it is only thesum of all gated tomograms that is reviewed in thismanner; this will alert the observer to most types ofgating errors due to arrhythmias manifested by anintermittent flashing of the images. However, for thedetection of gating errors due to some types oftransient arrhythmias, a full display of all count-versus-projection curve data is helpful. Cine reviewsoccasionally show abnormalities in the abdomen orthorax such as renal cysts or abnormal uptake thatmay be suspicious for neoplasm. The accuracy andclinical utility of these findings have not yet beenestablished.

5. Display–study review. It is strongly recommendedthat physicians use the active computer screen forreviewing images and use film and paper hard copiesonly for record-keeping purposes. Images producedby formatters onto transparency film or photographicpaper can have variable contrast, also termed gamma,and result in inconsistent image interpretation. Com-puter screen outputs are relatively more stable and
always have readily available monochromatic con-
trast bars or color code bars to the side of the images,enabling more consistent viewing conditions. In ad-dition, computer screens offer rapid sequential and/orcinematic displays of image data. For all of thesereasons, screen interpretation is strongly recom-mended over relying on interpretations from hardcopies.16

PERFUSION QUANTITATION

The display medium and translation table employedcan have a significant impact on image interpretation,going as far as to make a normal perfusion scan appearabnormal or vice versa. Quantitative analysis is a directway of measuring relative uptake of a perfusion tracerthat is independent of the display medium and translationtable and can thus greatly reduce variations in interpre-tation due to subjective analysis and inconsistent imagedisplay. Quantitative analysis also allows for the com-parison of a study with a gender-specific normal data-base. For this reason, quantitative analysis is recom-mended as part of image interpretation. Traditional,circumferential profile-based quantitative analysis con-sists of the following steps:

1. Image quantitation. This measures the relative ac-tivity, most often in a short-axis slice, by generating acircumferential profile also known as a radial plot.This is done by first identifying the center of thelumen and the inner and outer boundaries of theventricle. The activity in each part of the slice isdetermined by measuring the activity in each pixellying on a radius between the inner and outer bound-aries, analogous to moving along a spoke of a wagonwheel from the hub to the rim. The radius is thenrotated slightly (analogous to going to the next spokeof the wagon wheel), and the activity is again mea-sured. This is repeated until the entire circumferencehas been traversed. Determination of activity alongeach radius can be as simple as identifying themaximal pixel or as complicated as fitting the profileto a Gaussian curve.

2. Normalization and scaling. Myocardial perfusionimaging can only measure relative uptake. In order tocompare different studies or compare with normaldatabases, the images must be normalized to a certainvalue. Each slice can be normalized to its ownmaximum, but it is much more common to normalizeto the maximal pixel in the ventricle. All values aremultiplied by 100/value of the maximum pixel, whichsets the maximum pixel to 100 and all others to afraction of that. Finally, the ventricle is “scaled” to aconstant number of curves; smaller ventricles with
fewer short-axis slices are interpolated up, and larger


ventricles with more short-axis slices are decimateddown to achieve a constant number of curves. Thusrelative activity can be compared for any location intwo ventricles regardless of their absolute traceruptake or chamber size.

3. Polar plot generation (optional). Each radial plotgenerated in step 1 generates a series of numbers thatcan be displayed as a graph. The entire LV volumewould be represented by a series of graphs, which isdifficult to assimilate. Alternatively, these plots canbe used to make a “bulls-eye” or polar plot.17 Insteadof generating a graph, the radial profiles are used tomake a series of rings of activity, where the intensityof the ring corresponds to the relative activity of thecorresponding polar plot and the diameter of the ringsgrows moving toward the base of the heart. These canthen be fused into a single image where each succes-sive ring is formed around the one preceding it,analogous to the rings of a tree. This creates a2-dimensional image that reflects the relative activityin the 3-dimensional left ventricle with the apex at thecenter of the polar plot and the base as the outerring.17,18 The width of the rings can be adjusted toreflect the relative size of the corresponding short-axisimages so that a given area of the polar plot corre-sponds to a constant fraction of the volume of the leftventricle in a process known as “volume weighting.”

4. Database construction and analysis. It is frequentlydifficult to differentiate true perfusion defects fromsoft-tissue attenuation; women tend to have anteriordefects from breast attenuation, and men tend to haveinferior wall defects from diaphragmatic attenua-tion.19 A range for “normal” soft-tissue attenuationcan be defined by creating a normal database; gender-specific normal databases are usually employed due tothe different attenuation patterns for men and women.These can be generated by performing radial plotanalysis on “normals”—that is, patients proven not tohave coronary disease or, more often, patients with anacceptably low probability of having coronary disease(usually �5% or �1%). The mean and standarddeviation of activity for each point of the ventricle arecalculated for the male and female normals; the resultis the normal database for each gender. The normal-ized and scaled activity for each point of the ventricleof a patient is compared to the mean of the corre-sponding gender-based normal database; if the activ-ity is more than a predetermined number of standarddeviations below the mean of the normals (2.5 stan-dard deviations is most frequently used), it is consid-ered abnormal. It should be realized that this defini-tion of normal is statistical, not absolute. There will
still be overlap between normal and abnormal uptake.
5. Parameters. Different parametric images can begenerated to show the results of quantitative analysis,such as “blackout,” “severity,” and “reversibility”maps. A blackout map generates a polar plot for thepatient marking those points that fall a predeterminednumber of standard deviations below the gender-based normal limits and thus demonstrates the size orextent of the perfusion defect. Another type of para-metric image can be generated which shows thenumber of standard deviations that activity fallsbelow the mean of normals and thus demonstratesseverity. A third can be generated which quantitatesthe activity on the stress and rest images and demon-strates which areas show improved perfusion at rest,which is known as a reversibility map.

GATED SPECT

Acquisition. The introduction of technetium-basedperfusion tracers has resulted in images with sufficientcount density to allow for cardiac gating adding param-eters of wall motion, wall thickening, and EF to myo-cardial perfusion imaging.20-23 Gating requires a stableand consistent heart rhythm as well as sufficient temporalresolution to correctly characterize the cardiac cycle. Astable heart rate and rhythm can be achieved by rejectingheartbeats that fall out of range at the expense of anincrease in image time. This “beat length acceptancewindow” can vary from 20% to 100% of the expectedR-to-R duration, the recommended value being 20% if an“extra frame” is provided that allows the accumulation ofrejected counts. Most laboratories gate the heart for 8frames per cycle, although an increasing number oflaboratories have reported good results with 16 framesper cycle and have used the increased temporal samplingto derive more accurate estimates of LVEF as well asparameters of diastolic function. For either 8- or 16-frame gating, the recommendations are to avoid beatrejection. The lower count statistics achieved with Tl-201 imaging make gating more challenging with thisisotope, but many laboratories have reported satisfactoryresults using 8-frame gating in selected patients.24 Pro-cessing is done using commercially available software.

INTERPRETATION AND REPORTING

General Comments

The interpretation of myocardial perfusion SPECTimages should be performed in a systematic fashion toinclude (1) repeat evaluation of the raw tomographicimages to determine the presence of technical sources ofabnormalities and extracardiac activity; (2) interpretation
of images with respect to the location, size, severity, and


reversibility of defects as well as chamber sizes and, forTl-201, presence or absence of pulmonary uptake; (3)incorporation of the results of quantitative analysis; (4)consideration of functional data obtained from the study;and (5) consideration of clinical factors that may haveinfluenced the presence of any findings. All of thosefactors contribute to the production of a final clinicalreport. Guidelines for interpreting and reporting myocar-dial perfusion SPECT are listed in Tables 8 and 9.

Display

1. Recommended medium for display. It is stronglysuggested that the reading physician use the com-puter monitor screen rather than hard copy to inter-pret the study since a monitor screen is capable ofdisplaying more variations in gray scale or color(making it easier to discern smaller variations inuptake) and is more consistent than film. Moreover,it is not possible to properly view moving images(eg, raw tomographic data or gated images) on hardcopy. A linear gray-scale translation table is gener-ally preferred to color tables since it demonstratesconsistent grades of uptake compared to pseudocon-touring seen with color scales, but this is alsodependent on how familiar the reader is with a giventranslation table.16,25 The reader should be awarethat the appearance of an image can change signif-icantly when changing from one translation table toanother. A linear scale is preferred to nonlinear (eg,sigmoidal) scales since it most faithfully character-izes uptake over the range of activity. A logarithmicscale may be used for evaluating regions of lower-count density such as soft-tissue uptake and the rightventricle but should never be used for interpretingLV uptake.16,25

2. Conventional slice display of SPECT images.Three sets of images should be displayed: (a) a viewgenerated by slicing perpendicular to the long axis ofthe left ventricle (short axis); (b) a view of long-axistomograms generated by slicing in the vertical plane(vertical long axis); and (c) a view of long-axistomograms generated by slicing in the horizontalplane (horizontal long axis). The short-axis tomo-grams should be displayed with the apical slices tothe far left with progression of slices toward the basein a left-to-right fashion. The vertical long axisshould be displayed with septal slices on the left andprogression through the midventricular slices to thelateral slices in a left-to-right fashion. Similarly, thehorizontal long-axis tomographic display shouldproceed left to right from the inferior to the superior(anterior) surface. It is also recommended that, for
purposes of interpretation and comparison of se-
quential images (eg, stress and rest, rest and redis-tribution), these images be displayed aligned andadjacent to each other serially. There are two widelyused approaches to image normalization. Each series(vertical, horizontal, short axis) may be normalizedto the brightest pixel in the entire image set, which isknown as “series normalization.” This is consideredto provide the most intuitively easy way to evaluatethe extent and severity of perfusion defects. Thedrawbacks of this approach are its sensitivity to focalhot spots, the frequently poor visualization of nor-mal structures at the base and apex of the leftventricle, and the lack of an ideal display of eachindividual slice.

The other approach is “frame normalization” inwhich each image is normalized to the brightestpixel within the frame. That method provides opti-mal image quality of each slice. The drawback ofthis approach is that the brightness of each slice isunrelated to the peak myocardial activity in theentire series such that gradations in activity betweenslices of a series may be lost. That drawback ismitigated by the display of three orthogonal planes.

3. Three-dimensional display. Most commercial soft-ware programs allow creation of 3-dimensional dis-plays. These help less experienced readers identifycoronary distributions associated with perfusion de-fects. These should be used only as an adjunct to, nota replacement for, the conventional image format-ting described above.

Evaluation of Images for Technical Sourcesof Error

4. Patient motion. The interpreting physician shouldalso review the raw tomographic images for possiblesources of artifact. Images should again be inspectedfor the presence of patient motion. A cine display ofthe planar projection data is highly recommendedbecause motion in both the craniocaudal and hori-zontal axes is readily detectable. Additionally, astatic “sinogram” and sometimes “linogram” may beused to detect motion. Software routines are avail-able for quantitation and correction of motion. Theexperienced reader should be familiar with thenormal appearance of raw tomograms and be able toidentify motion artifact. In patients who have had atechnetium-based perfusion agent, considerationshould be made for repeating the image acquisitionwhere feasible when significant motion is detected.Alternatives such as planar imaging or prone posi-tioning may be considered as well. The effect ofpatient motion on the final reconstructions is com-
plex.6,26-28 Generally, up-and-down motion (espe-


cially when the heart returns to the same baseline)has less of an effect on the accuracy of the study thansideways motion. Also, up-and-down motion ismuch easier to correct either manually or withsemiautomated software. Rotation currently cannotbe corrected either manually or with available mo-tion correction software. Since motion correctionsoftware may sometimes introduce motion artifact,corrected raw tomographic images should be evalu-ated in the same way for adequacy of the correction.

5. Attenuation and attenuation correction. The cinedisplay of the planar projection images is alsorecommended for the identification of sources ofattenuation, the most common being diaphragmaticin men and the breast in women.29 Breast attenuationartifact is most problematic when it is differentbetween the rest and stress images. When breastattenuation artifact is more prominent on the stressimages than on the rest, it can be very challenging toexclude ischemia. Breast attenuation can sometimesbe improved by repeating the acquisition with thebreast repositioned. Diaphragmatic attenuation andbreast attenuation may be reduced by imaging thepatient prone. Hardware and software for attenuationand scatter correction are commercially availableand may obviate or at least mitigate these commonattenuation artifacts. The evaluation of attenuation-corrected (AC) images is performed with the sameapproach as that used for non-AC images. As withthe interpretation of non-AC studies, it is essentialthat the interpreting physician be familiar with thesegment-by-segment normal variation of uptake ofradioactivity at stress and rest associated with thespecific attenuation correction system that is beingused.30-32 AC images are displayed in the samemanner as uncorrected images. Because the cur-rently available correction algorithms are imperfect,it is recommended that the uncorrected data beinterpreted along with the AC data.

6. Reconstruction artifacts. Superimposed bowelloops or liver activity may create artifactually in-tense uptake in adjacent myocardium that couldmask a real perfusion defect or be misinterpreted asreduced uptake in adjacent or contralateral seg-ments. Non-superimposed but adjacent extracardiacactivity may also affect the reconstructed myocardialimages. Intense activity in bowel loops or adjacentliver may cause a negative reconstruction artifact,resulting in an apparent reduction in activity in theadjacent myocardial segments. There is currently noreliable correction for such artifacts, although theymay be less prominent with iterative as opposed to
filtered backprojection techniques. They can often
be eliminated by repeating the acquisition after theactivity level in the adjacent structure has decreased.

7. Myocardial statistics. Many factors are involved inthe final count density of perfusion images includingbody habitus, exercise level, radiopharmaceuticaldose, acquisition time, energy window, and collima-tion. The interpreting physician should make note ofthe count density in the planar projection imagesbecause the quality of the reconstructed data is adirect reflection of the raw data. Perfusion defectscan be artifactually created simply because of poorstatistics. As a general rule, peak pixel activity in theLV myocardium in an anterior planar projectionshould exceed 100 counts for a Tl-201 study and 200counts in a Tc-99m study.

Initial Image Analysis and Interpretation

The initial interpretation of the perfusion study shouldbe performed without any clinical information other thanthe patient’s gender, height and weight, and peak exerciseheart rate. Such an approach minimizes the bias in studyinterpretation. All relevant clinical data should be reviewedafter a preliminary impression is formed.

8. Ventricular dilation. Before segmental analysis ofmyocardial perfusion, the reader should note whetherthere is LV enlargement at rest or during stress.Dilation on both the stress and resting studies usuallyindicates LV dysfunction, although it may be seen involume overload states with normal ventricular func-tion. Transient ischemic dilation (TID) has been de-scribed as a marker for high-risk coronary disease.33 Itis most likely due to diffuse subendocardial isch-emia34,35 and can be seen in other conditions, such asmicrovascular disease, that cause diffuse subendocar-dial ischemia even in the absence of epicardial disease.TID is typically described quantitatively but may bequantified.36 Normal limits by quantitation will dependon both the software and perfusion agents being used.

9. Lung uptake. The presence of increased lung up-take after thallium perfusion imaging has beendescribed as an indicator of poor prognosis andshould therefore be evaluated in all patients whenusing this perfusion agent.34,35,37 No clear consensushas emerged as to the significance of lung uptakewith technetium-based perfusion agents.

10. Right ventricular uptake. Right ventricular (RV)uptake may be qualitatively assessed on the rawprojection data and on the reconstructed data. Thereare no established quantitative criteria for RV up-take, but in general, the intensity of the right ventri-cle is approximately 50% of peak LV intensity. RV
uptake increases in the presence of RV hypertrophy,


most typically because of pulmonary hypertension.38

The intensity of the right ventricle may also appearrelatively increased when LV uptake is globallyreduced.39-41 Regional abnormalities of RV uptakemay be a sign of ischemia or infarct in the distribu-tion of the right coronary artery. The size of the rightventricle should be noted.

11. Noncardiac findings. Both thallium- and techne-tium-based agents can be concentrated in tumors,and uptake outside the myocardium may reflectunexpected pathology. However, the accuracy and,in particular, the specificity of myocardial perfusionimaging for diagnosing noncardiac conditions havenot been established. Splanchnic Tl-201 activityfollowing adequate exercise stress (�85% maxi-mum predicted heart rate) is generally reducedcompared to resting images. This difference is notpresent following pharmacologic Tl-201 stress test-ing with dipyridamole, adenosine, or dobutamine.

12. Perfusion defect location. Myocardial perfusiondefects should be identified by use of visual analysisof the reconstructed slices. The perfusion defectsshould be characterized by their location as theyrelate to specific myocardial walls—that is, apical,anterior, inferior, and lateral. The term posteriorshould probably be avoided because it has beenvariably assigned either to the lateral wall (circum-flex distribution) or to the basal inferior wall (rightcoronary distribution) and is thus ambiguous. Stan-dardization of segment nomenclature is highly rec-ommended. (See the segmentation models describedbelow.)

13. Perfusion defect severity and extent: Qualitative.Defect severity is typically expressed qualitativelyas mild, moderate, or severe. Mild defects may beidentified by a decrease in counts compared toadjacent activity without the appearance of wallthinning, moderate defects demonstrate wall thin-ning, and severe defects are those that approachbackground activity.42,43 Defect extent may be qual-itatively described as small, medium, or large. Insemiquantitative terms, small represents 5% to 10%,medium represents 15% to 20%, and large repre-sents 20% of the left ventricle or greater.43 Alterna-tively, defect extent may also be estimated as afraction such as the “basal one half” or “apical onethird” of a particular wall or as extending from baseto apex. Defects whose severity and extent do notchange between image sets (eg, stress and rest) aretypically categorized as “fixed” or nonreversible.When changes do occur, a qualitative description ofthe degree of reversibility is required.

14. Perfusion defect severity and extent: Semiquan-
titative. In addition to the qualitative evaluation of
perfusion defects, it is preferred that the physicianmay also apply a semiquantitative method on thebasis of a validated segmental scoring system. Thisapproach standardizes the visual interpretation ofscans, reduces the likelihood of overlooking signif-icant defects, and provides an important semiquan-titative index that is applicable to diagnostic andprognostic assessments. It is generally consideredpreferable to use a system with at least 16 segments.

The quality assurance (QA) committee of the Amer-ican Society of Nuclear Cardiology has considered sev-eral models for segmentation of the perfusion images andhas previously recommended either a 17- or 20-segmentmodel for semiquantitative visual analysis. The modelsuse three short-axis slices (apical, mid, and basal) torepresent most of the ventricle and one vertical long-axisslice to better represent the LV apex. In both the 17- and20-segment models, the basal and mid short-axis slicesare divided into 6 segments. In the 17-segment model,the apical short-axis slice is divided into 4 segments,whereas in the 20-segment model, the apical short-axisslice is divided into 6 segments. In the 17-segmentmodel, a single apical segment is taken from the verticallong-axis slice, whereas in the 20-segment model, theapex is represented by 2 segments. Each segment has aspecific name. In order to facilitate consistency ofnomenclature with other imaging modalities, the 17-segment model has become the preferred nomenclature.

Seventeen-segment nomenclature (Figure1). Segments 1, 7, and 13 represent the basal (1), mid(7), and apical (13) anterior segments. Segments 4, 10,

Figure 1. SPECT myocardial perfusion imaging: 17-segmentmodel.

and 15 represent the basal (4), mid (10), and apical (15)


inferior segments. The septum contains 5 segments, thebasal anteroseptal (2), the basal inferoseptal (3), the midanteroseptal (8), the mid inferoseptal (9), and the apicalseptal (14). Similarly, the lateral wall is divided into thebasal anterolateral (6), the basal inferolateral (5), the midanterolateral (12), the mid inferolateral (11), and theapical lateral (16). The long-axis apical segment is calledthe apex.

Twenty-segment nomenclature (Figure 2). Seg-ments 1, 7, and 13 represent the basal (1), mid (7), andapical (13) anterior segments. Segments 4, 10, and 16represent the basal (4), mid (10), and apical (16) inferiorsegments. The septum contains 6 segments, the basal (2),mid (8), and apical (14) anteroseptal and the basal (3),mid (9), and apical (15) inferoseptal. Similarly, thelateral wall contains 6 segments, the basal (6), mid (12),and apical (18) anterolateral and the basal (5), mid (11),and apical (17) inferolateral segments. The apex from thevertical long-axis slice is divided into anteroapical (19)and inferoapical (20) segments.

The myocardial segments may be roughly assignedto coronary arterial territories as indicated in Figure 3 aslong as the reader realizes that there can be considerablevariation among patients especially in the inferior andinferolateral segments of the left ventricle due to thevariable extent of the circumflex and right coronaryartery territories.

Semiquantitative Scoring System: The Five-PointModel

The use of a scoring system provides a reproducible

Figure 2. SPECT myocardial perfusion imaging: 20-segmentmodel.

semiquantitative assessment of defect severity and ex-

tent. A consistent approach to defect severity and extentis clinically important because both of those variablescontain independent prognostic power. Furthermore,semiquantitative scoring can be used to more reproduc-ibly and objectively designate segments as normal orabnormal. Points are assigned to each segment in directproportion to the perceived count density of the segment(Table 7).

In addition to individual scores, it has been recom-mended that summed scores be calculated. The summedstress score equals the sum of the stress scores of all thesegments, and the summed rest score equals the sum ofthe resting scores or redistribution scores of all thesegments. The summed difference score equals thedifference between the summed stress and the summedresting (redistribution) scores and is a measure of revers-ibility. In particular, the summed stress score has been

Figure 3. SPECT myocardial perfusion imaging: coronaryartery territories. LAD, Left anterior descending artery; RCA,right coronary artery; LCX, left circumflex artery.

Table 7. Semiquantitative scoring system: 5-pointmodel

Category Score

Normal perfusion 0Mild reduction in counts—not definitely

abnormal1

Moderate reduction in counts—definitelyabnormal

2

Severe reduction in counts 3Absent uptake 4

shown to have significant prognostic power.43 Before


scoring, it is necessary for the interpreting physician tobe familiar with the normal regional variation in countdistribution of myocardial perfusion SPECT.

15. Perfusion defect severity and extent: Quantita-tive. Quantitative analysis is useful to supplementvisual interpretation.42-44 Most techniques of quan-titative analysis are based on radial plots of short-axis slices. Different techniques analyze the apexseparately. These plots are then normalized to allowcreation of or comparison to normal databases.Defects can be defined as where activity falls a givenamount below the mean of a normal database toevaluate size and severity of defects. Quantitation ofthe stress images is compared to the rest images toassess the degree of ischemia versus infarction. It iscustomary to generate separate normal databasesbased on gender as well as the perfusion agent used.This quantitative analysis is usually displayed as a“bulls-eye” or polar plot.45 The quantitative pro-grams are effective in providing an objective inter-pretation that is inherently more reproducible thanvisual analysis, eliminates the variability of theappearance of a defect when viewed in differentmedia (with different gammas) and different trans-lation tables, and is particularly helpful in describingchanges between two studies in the same patient.Quantitative analysis also serves as a guide for theless experienced observer who may be uncertainabout normal variations in uptake. Quantitative pro-grams are by no means sophisticated enough tounequivocally differentiate perfusion defects fromartifact but help in understanding the range of uptakethat can be encountered in patients without disease.Because of artifacts during imaging and also thenature of coronary blood flow, there will always bean overlap between normals and patients with mildperfusion defects; this overlap can be reduced butnot completely eliminated by careful attention toimage acquisition and reconstruction. Thereforequantitative analysis should only be used as anadjunct to and not a substitute for visual analysis.

Defect extent may be quantitatively expressed as apercentage of the entire left ventricle or as a percent-age of individual vascular territories, the latter beingless reliable because of the normal variations incoronary anatomy. Defect severity may be quantita-tively expressed as the number of standard devia-tions by which the segment varies from the normalrange for that particular segment or segments. De-fect reversibility may also be expressed as a percent-age of the entire left ventricle or of a vascular
territory.
16. Reversibility. Reversibility of perfusion defectsmay be categorized qualitatively as partial or com-plete, the latter being present when the activity in thedefect returns to a level comparable to surroundingnormal myocardium. The semiquantitative scoringsystem may be used to define reversibility as a�2-grade improvement or improvement to a scoreof 1. Reversibility on a quantitative polar plot or on3-dimensional displays will depend on the particularsoftware routine in use and the normal referencedatabases used in the program. Areas of reversibilityare typically described by pixels that improve to lessthan 2.5 SDs from the normal reference redistribu-tion or resting database. How many pixels have toimprove for reversibility to be deemed present isarbitrary.

So-called reverse redistribution may be seen instress delayed thallium imaging sequences and hasbeen described in rest delayed technetium sestamibisequences. Reverse redistribution refers to segmentswith decreased or normal intensity on the initial setof images that show even less relative intensity onthe delayed images. The interpretation of the findingremains controversial, but in certain clinical situa-tions, it seems to represent segments with a mixtureof viable and nonviable myocardium that are fre-quently supplied by patent infarct-related arteries.46

GATED MYOCARDIAL PERFUSION SPECT

Because of the comparatively low additional costand substantial benefit of the information obtained, gatedstudies of ventricular function should be a routine part ofmyocardial perfusion SPECT.20 A systematic approachto display and interpretation of the ventricular functionderived from gated SPECT is important.

17. Gated SPECT display. Multiple ventricular slicesshould be evaluated. At a minimum, a quad-screendisplay of apical and mid-basal short-axis, a mid-ventricular horizontal long-axis, and a mid-ventric-ular vertical long-axis slice should be viewed. Otherslices may be viewed for completeness or to resolvea discrepancy between the clinical impression andwhat is seen on the four standard cine views. Ideally,the software should allow the user to scroll throughany of the slices in any axis in cine mode. Each viewshould be normalized to the series of end-diastolic toend-systolic slices to maintain the count densitychanges that reflect wall thickening. When available,software routines that automatically define epicar-dial and endocardial borders and that subsequentlycalculate ventricular volumes and EF should be
applied.


Table 8. Myocardial perfusion SPECT: Guidelines for interpretation


A. Display1. Medium

a. Computer screen Preferred 1b. Film hard copy Discouraged 1

2. Formata. Conventional slice display Preferred 2

i. Frame normalization Optional 2ii. Series normalization Preferred 2

b. Three-dimensional display Optional 3B. Technical sources of error

1. Motion Standard 42. Attenuation Standard 5

a. Attenuation correction Optional 53. Reconstruction artifacts Standard 64. Myocardial statistics Standard 7

C. Initial image interpretation1. Ventricular dilation

a. Qualitative Standard 8b. Quantitative Optional 8

2. Lung uptakea. Qualitative Standard 9b. Quantitative Preferred 9

3. Non-cardiac Standard 114. Perfusion defect assessment

a. Location Standard 12b. Extent/severity

i. Qualitative Standard 13ii. Semiquantitative Optional 14iii. Quantitative Optional 15

5. Reversibility Standard 16D. Gated SPECT

1. Display Standard 172. QC Standard 183. Regional wall motion Standard 194. Regional wall thickening Standard 195. LVEF

a. Qualitative Standard 20b. Quantitative Preferred 20

6. LV volumea. Qualitative Standard 20b. Quantitative Recommended 20

E. Integration of perfusion and function results Standard 21F. Myocardial viability

1. Qualitative Standard 222. Semiquantitative Optional 233. Quantitative Preferred 24

G. Modification of interpretation Preferred 25


Table 9. Myocardial perfusion SPECT: Guideline for reporting


A. Demographic data1. Name Standard 262. Gender Standard 263. Age Standard 264. Date(s) of acquisition(s) Standard 265. Medical record identification Standard 266. Height/weight (body surface area) Standard 267. Relevant medications Optional 268. Indication for study Standard 28

B. Acquisition parameters1. Type(s) of studies Standard 272. Radionuclide(s) and doses Standard 27

C. Results: Exercise/intervention data1. Resting ECG findings Standard 292. Exercise/intervention parameters

a. Heart rate, blood pressure, % maximal predicted heart rate, metabolicequivalents

Standard 30

b. Symptoms Standard 30c. Reason for terminating Standard 30d. ECG changes with exercise Standard 30

D. Results: perfusion scan data1. Potential sources of error

a. Motion Standard 4b. Attenuation Standard 5c. Adjacent/overlapping uptake Standard 6

2. Chamber sizes Standard 83. Lung uptake (thallium)

a. Qualitative Standard 9b. Quantitative Preferred 9

4. Initial defect location Standard 125. Initial defect severity and extent

a. Qualitative Standard 13b. Semiquantitative Preferred 14c. Quantitative Optional 15

6. Reversibilitya. Qualitative Standard 16b. Semiquantitative Preferred 16c. Quantitative Optional 16

7. RV uptake Standard 108. Abnormal noncardiac uptake Standard 11

E. Results: gated SPECT1. Regional wall motion Standard 192. Regional wall thickening Standard 193. EF

a. Qualitative Recommended 20b. Quantitative Recommended 20

4. LV volume Optional 20
F. Overall study quality Optional 31


Regional wall motion should be interpreted with agray-scale display. When computer edge analysissoftware is available, the physician may choose toanalyze wall motion by use of the assigned endocar-dial and epicardial contours, but reference shouldalso be made to the wall motion without computer-derived edges. Regional wall thickening may beanalyzed in gray scale or in a suitable color scheme,although color displays may make it easier to seechanges in count intensity.

18. Gated SPECT QC. All the QA procedures forroutine SPECT are applicable to gated SPECT withthe addition of the evaluation of the adequacy of theelectrocardiographic (ECG) gate.21 The most com-mon manifestation of poor gating is the appearanceof a flashing pattern on the rotating planar projectionimages that results from count loss in the laterframes. Ideally, a heart rate histogram should also beviewed to verify beat length uniformity. Inspectingthe time-volume curve is particularly useful sincegating errors may distort the curve. As yet, there isno clear consensus on the beat length window forgated SPECT acquisitions. As in blood pool imag-ing, the narrower the window, the more physiologicthe data, but this runs the risk of compromising thequality of the SPECT perfusion images unless theacquisition software allows for an “extra frame” toaccumulate rejected counts during the gated acqui-sition. Another important aspect of QC is a visual orquantitative determination that the number of countsacquired in each frame of the gated study wasadequate for assessment of function. Software thatcollects all counts into a separate bin for the summedimage can minimize the effect that gating errorshave on the summed image.

19. Gated SPECT: Regional wall motion and thick-ening. Regional wall motion should be analyzed byuse of standard nomenclature: normal, hypokinesis,akinesis, and dyskinesis. Hypokinesis may be furtherqualified as mild, moderate, or severe. A semiquan-

Table 9. (Continued)

G. Conclusion1. Normal/abnormal

a. Three categoriesb. Five categories

2. Probability of CAD3. Estimated risk of adverse events

titative scoring system is recommended, where 0 is

normal, 1 is mild hypokinesis, 2 is moderate hypo-kinesis, 3 is severe hypokinesis, 4 is akinesis, and 5is dyskinesis.

This is comparable to the 5-point scoring systemused in both contrast and radionuclide ventriculog-raphy. As in any assessment of regional ventricularfunction, one must be cognizant of expected normaland abnormal variations such as the reduced wallexcursion at the base compared with the apex, thegreater excursion of the basal lateral wall comparedwith the basal septum, and paradoxical septal mo-tion, which may result from left bundle branchblock, post pericardiotomy, or pacing from the rightventricle.

Normal myocardial wall thickness is below theresolution of image reconstruction from currentlyavailable SPECT systems. However, regional wallthickening can be estimated by use of the countincrease from end diastole to end systole. Visually, itis not as easy to assign degrees of abnormality ofthickening as it is to wall motion. However, theevaluation of thickening with gated perfusionSPECT lends itself to quantitation because it ischaracterized by count changes.

Wall motion and wall thickening are generallyconcordant. The principal exception to this occurs inpatients who have undergone cardiac surgery whereseptal wall motion is frequently abnormal (paradox-ical) but there is normal wall thickening. Rather thanseparately scoring wall motion and wall thickening,it is commonly accepted to incorporate the twofindings into a single score while noting the presenceof discordance in wall motion and wall thickeningwhen it occurs. In addition to noting LV wallmotion, wall thickening, and EF, the size of the leftand right ventricles should be observed, and thefunction of the right ventricle should be noted.

Quantitative normal databases are now availablefor assessment of regional wall thickening.

20. LVEF and volume. LVEF and LV and RV chamber


Recommended 16Optional 16Optional 32Optional 32

sizes should routinely be evaluated qualitatively.47


EF may be categorized as normal or mildly, moder-ately, or severely reduced.22 Volume may be cate-gorized as normal or mildly, moderately, or severelyincreased. Alternatively, LVEF and end-diastolicand end-systolic volumes may be calculated withgeometric models applied to the reconstructed dataset. Several software routines that correlate wellwith contrast and other radionuclide measurementsare commercially available.22

21. Integration of perfusion and function results. Theresults of the perfusion and gated SPECT data setsshould be integrated into a final interpretation. Thewall motion is particularly helpful in distinguishingreal nonreversible perfusion defects from attenuationor motion artifacts. Fixed perfusion defects that donot show a corresponding abnormality of motion orthickening are more likely to be due to artifacts,especially if the clinical data do not support priorinfarction.29

Myocardial Viability

22. Viability: Qualitative assessment. The assessmentof myocardial viability is a complex issue made evenmore difficult by the lack of consensus in theliterature of the precise meaning of the term viabil-ity—whether it refers merely to the absence of scaror requires improvement in wall motion after revas-cularization. It is, however, clear that the quantita-tive uptake of radionuclides such as Tl-201 and theavailable technetium agents does correlate withmyocardial viability as defined by post-revascular-ization improvement in both tracer uptake and re-gional function. Myocardial segments with normalor mildly reduced tracer uptake at rest or on delayedimaging almost invariably prove to be viable. Themajority of myocardial segments in which there isunequivocal improvement of uptake on either redis-tribution or resting images also prove to be viable.The more difficult challenge for the single photonassessment of viability is in segments with severelyreduced tracer uptake.

23. Myocardial viability: Semiquantitative assess-ment. The semiquantitative scoring system de-scribed above may be used to assess viability asfollows. Segments with rest, reinjection, or redistri-bution scores of 0 (normal perfusion) and 1 (slightreduction in counts) are considered viable. Segmentswith rest, redistribution, or reinjection scores of 2(moderately decreased perfusion) are consistent witha combination of viable and nonviable myocardium,and segments with scores of 3 and 4 are generallynonviable. Segments with final scores of 4 are
considered nonviable.
24. Myocardial viability: Quantitative assessment.An alternate and perhaps more rigorous approach tothe assessment of the viability of any segment is thequantitative determination of ischemic-to-normal ra-tios. Regions of interest may be placed over thesegment in question and over the most normalsegment of the myocardium in that particular seriesof images. The analysis should be applied to theresting images for technetium images or to theresting, redistribution, or reinjection images for thal-lium. When this approach is used, one must take intoaccount the normal count variations such as therelatively reduced counts in the normal inferior wall.Segments with ratios of less than 0.30 are considerednonviable. Areas with ratios greater than 0.50 areconsidered viable, whereas areas with ratios of 0.30to 0.50 are equivocal for viability. As indicatedabove for the semiquantitative approach, such re-gions require additional data such as wall motion ofthe region, exercise perfusion in the region, thechange in perfusion or wall motion after nitroglyc-erin, the response of regional function to low-dosedobutamine, or myocardial metabolic imaging withfluorine 18 fluorodeoxyglucose.22

It is also important to recognize that viability of agiven segment does not necessarily equate to im-provement in clinical outcome after revasculariza-tion unless enough segments that are viable areavailable. The critical number of segments necessaryto justify revascularization strategies has not beenadequately studied.

Modification of Interpretation by RelevantClinical Information

25. Due to imperfections in the technology as well as thegradual impairment of coronary blood flow as ste-noses become hemodynamically significant, therewill always be overlap between normal and mildlyabnormal perfusion scans. In these patients it is oftenparticularly helpful to incorporate other clinicalinformation (eg, symptoms, risk factors, ST-segmentchanges, exercise tolerance) as well as prognosticinformation in order to help the referring physicianmake the most appropriate management decisionsfor patients. Homogeneous perfusion images of pa-tients who have other markers of severe ischemia,such as marked ST-segment changes, should becarefully evaluated for adjunctive markers of isch-emia such as TID or increased lung uptake (withthallium) in order to identify those patients withbalanced ischemia. The majority of artifacts encoun-tered will produce mild defects; therefore moderate
or severe defects, in the absence of dramatic artifact,


should be considered as reflecting pathology. Fi-nally, it needs to be understood that not all pathologydetected by perfusion imaging reflects epicardialcoronary artery disease (CAD).

Reporting of SPECT Myocardial Perfusion ScanResults

26. Subject information. The age, gender, height,weight, and body surface area should be included inthe report because they may directly affect the imageresults and interpretation. For medical records pur-poses, any identification number should be included.Pertinent medications that may influence the resultsmay be included.

27. Type of study. The imaging protocol should bespecified, including the radiopharmaceutical anddose, imaging technique (gated vs ungated, supine orprone), imaging sequence (stress/4-hour redistribu-tion, 1-day or 2-day rest/stress or stress/rest, and soon), and a specific statement about whether imageswere or were not corrected for attenuation. Thedate(s) of study acquisitions should also be included.

28. Indication for study. Placing the indication for thestudy in the report helps focus the interpretingphysician on the clinical question raised by theordering physician and may be subsequently impor-tant for reimbursement issues.

29. Resting ECG findings. Inclusion of ECG findingsthat may have a direct bearing on the study interpre-tation should be included such as the presence of leftbundle branch block or LV hypertrophy.

30. Summary of stress data. The type of stress (bicycleor treadmill) and the protocol should be identified(Bruce, modified Bruce, Naughton, manual). Forpharmacologic stress, the agent, route of administra-tion, and dose should be indicated. The reason forstopping the test should be noted. All symptomsexperienced by the patient during stress (eg, chestpain, dyspnea, claudication, dizziness) should bementioned.

If a separate stress test report is generated, thenthe stress variables that could impact on the perfu-sion study quality or findings should be included inthe perfusion scan report. At a minimum, the reportshould include the total exercise duration, maximalheart rate and percent of predicted maximum heartrate, resting and maximal blood pressure achievedand workload achieved (estimated metabolic equiv-alents), and magnitude (in millimeters) and locationof any ST-segment deviation.

If only one report is used for both the exercise orpharmacologic study and the perfusion results, then
more detail about the stress test should be included
such as time of onset, duration and exact ECG leadswith ST-segment changes, the type of chest pain(typical, atypical, non-anginal) and its severity(mild, moderate, severe), and the presence of ar-rhythmia.

31. Overall study quality. Including a statement aboutthe quality of the study is helpful, as it alerts thephysicians using the report to any shortcomings thatmight reduce the accuracy of the data and theirinterpretation.

32. Conclusions. The final interpretation of the scanshould obviously reflect the reader’s impression asto whether the scan is normal or abnormal. This maybe refined to address uncertainty by adding 3 or 5categories reflecting certainty. That is, the scan maybe on a 2-category scale as normal or abnormal; a3-category scale as normal, equivocal, or abnormal;or a 5-category scale as normal, probably normal,equivocal, probably abnormal, or abnormal.

33. Diagnosis and prognosis of CAD. The probabilityof CAD may be determined with available algo-rithms that use the pre-scan likelihood of CAD asdetermined by age, gender, character of chest pain,the number of coronary risk factors, and the resultsof the stress electrocardiogram. The perfusion dataare then added to the model to produce a probabilityof CAD. A qualitative probability may be reportedon the basis of the definite presence or absence of aperfusion defect, the severity and extent of anyperfusion defects, and the presence of other markersof CAD such as transient LV dilation, post-stressstunning, or increased lung uptake. When CAD isknown to be present, the likelihood of stress-inducedischemia is reported instead of the likelihood ofsignificant CAD.

Though not widely available, some large labora-tories have enough internal follow-up data to be ableto statistically predict outcomes (death and nonfatalmyocardial infarction) on the basis of perfusionimage scores. If such data are available, incorpora-tion of the likelihood of an adverse event in thereport is desirable. Otherwise, a qualitative state-ment about risk may be appropriate because thelikelihood of an adverse event increases with thepresence of any of the following: perfusion defectsin multiple vascular territories, transient LV dilation,increased lung uptake, and decreased LV systolicfunction.

34. Clinical interpretation of AC SPECT studies. Theinterpretation of AC SPECT myocardial perfusionimages follows a similar approach to that used foruncorrected myocardial perfusion images, but thereare differences, and these should be taken into
account in order to obtain good results. The normal


distributions of perfusion tracer uptake are signifi-cantly different with AC compared to uncorrectedstudies, and because of this, it is important that theinterpreting physician have available and learn da-tabases of normal tracer distribution(s). Althoughfrom system to system these normal distributions aregenerally relatively similar, there can be differencesthat are dependent on the geometry of the imagingsystem, acquisition protocol, and processing algo-rithms.31 There can also be differences in normaldistribution(s) related to patient gender and ventric-ular volume. The interpreting physician must beaware of these differences if they exist for theirimaging system(s) and take them into account on apatient-by-patient basis when assessing clinical stud-ies.

AC SPECT studies generally have more uniformregional activity in the anterior, septal, inferior, andlateral walls, but mild reductions in apical and distalanterior activity are typical of the normal AC imagedistribution. This apical and distal anterior activityreduction is similar to that seen with positron emis-sion tomography myocardial perfusion imaging.This finding becomes more prominent when resolu-tion recovery and scatter correction are included inthe AC processing workflow and is often moreprominent in patients with larger hearts. In low-likelihood normal patients, this reduction in distalactivity is generally more prominent in men than inwomen, as men generally have larger hearts. If menand women with similar heart sizes are compared,the difference disappears. The unsuspecting ob-server may mistake this expected normal activityreduction for a distal or mid and distal left anteriordescending coronary perfusion defect. In general,the success of AC SPECT appears related to thediligence of the clinical laboratory in followingrecommended procedures for image acquisition, re-construction, QA, display, and quantification. Al-though it may seem reasonable that AC SPECTshould simply provide better images, like thosewithout correction, but with the bothersome artifactsthat often result in false-positive tests removed, thenormal distributions are significantly different andmust be accounted for to achieve optimal clinicalbenefit. The distribution of normal activity is differ-ent with AC SPECT. If these differences are notunderstood by interpreting physicians, AC SPECTwill be unreliable. Likewise, quantification and dis-play programs without appropriate normal AC data-bases should not be used for quantification, asspurious results will occur.

QA requirements are more demanding with AC
than non-AC images and should be carefully as-
sessed for each patient study. Artifacts due to move-ment, either respiration or patient movement, mis-registration, and extracardiac radiotracer uptake canbe amplified by the iterative algorithms that areemployed in AC reconstructions and processing.The quality and registration of the attenuation maps(or mu maps) with the emission image data areadditional key factors that must be ensured, and ifthey cannot be ensured, the associated AC imagesshould be read with greater caution. QA tools to aidthese assessments of registration and mu map qualityare still not uniformly available, but this shouldimprove in the near future.

For the clinical interpretation of AC SPECTmyocardial perfusion images, it is recommendedthat AC and non-AC images be displayed side byside with displays of the normal activity distribu-tion(s) and their variance distribution available asrequired for comparison. This requires the avail-ability of normal databases specific for the imag-ing device, imaging protocol, and processing ap-proach employed clinically. Extracardiac activityespecially when combined with respiratory and/orpatient movement can introduce artifacts and/ornormalization errors that may require renormal-ization or abandonment of the AC images alto-gether. Artifactual reductions in activity mostoften affecting the apparent anterior and/or lateralwall perfusion tracer uptake can occur when thereis misregistration of SPECT and mu map imagessuch that the myocardial activity from the SPECTimages is matched with the relatively low attenu-ation coefficients for adjacent lung tissue in themu maps. Some attenuation correction– capableSPECT systems acquire the SPECT and the trans-mission image data sequentially rather than simul-taneously. If there is a change in position ofarm(s) or breasts between emission and transmis-sion imaging, even though there may be perfectregistration of the heart in the emission andtransmission images, there can be artifactual de-fects introduced into the SPECT perfusion imagesby the misregistration of tissues outside thethorax.

The SPECT/CT systems that have become avail-able recently require sequential emission and trans-mission imaging with a change in bed positionbetween acquisitions. Although the quality of the mumaps with these systems will consistently far exceedthe quality of sealed-source transmission system mumaps, the greatly improved resolution of the mumaps they provide make it even more important thatregistration of emission and transmission reconstruc-
tions be exact to a tolerance of less than 1 pixel.


Acknowledgment

We acknowledge the excellent editorial assistance ofPatricia Upchurch, Director of Quality Assurance, AmericanSociety of Nuclear Cardiology.

References

1. Segall GM, Davis MJ. Prone versus supine thallium myocardialSPECT: a method to decrease artifactual inferior wall defects.J Nucl Med 1989;30:548-55.

2. Kiat H, Van Train KF, Friedman JD, Germano G, Silagan G, WangFP, et al. Quantitative stress-redistribution thallium-201 SPECTusing prone imaging: methodologic development and validation.J Nucl Med 1992;33:1509-15.

3. Esquerre JP, Coca FJ, Martinez SJ, Guiraud RF. Prone decubitus:a solution to inferior wall attenuation in thallium-201 myocardialtomography. J Nucl Med 1989;30:398-401.

4. Nishina H, Slomka PJ, Abidov A, et al. Combined supine andprone quantitative myocardial perfusion SPECT: method develop-ment and clinical validation in patients with no known coronaryartery disease. J Nucl Med 2006;47:51-8.

5. Slomka PJ, Nishina H, Abidov A, et al. Combined quantitativesupine-prone myocardial perfusion SPECT improves detection ofcoronary artery disease and normalcy rates in women. J NuclCardiol 207;14:44-52.

6. Friedman J, Van Train K, Maddahi J, Rozanski A, Prigent F,Bietendorf J, et al. “Upward creep” of the heart: a frequent sourceof false-positive reversible defects during thallium-201 stress-redistribution SPECT. J Nucl Med 1989;30:1718-22.

7. Eisner RL, Nowak DJ, Pettigrew R, Fajman W. Fundamentals of180 degree acquisition and reconstruction in SPECT imaging.J Nucl Med 1986;27:1717-28.

8. Liu YH, Lam PT, Sinusas AJ, Wackers FJ. Differential effect of180 degree acquisition and reconstruction in SPECT imaging.J Nucl Med 2002;43:1115-24.

9. Maniawski PJ, Morgan HT, Wackers FTH. Orbit-related variationin spatial resolution as a source of artifactual defects in thallium-201 SPECT. J Nucl Med 1991;32:871-5.

10. Germano G, Kiat H, Kavanagh PB, Moriel M, Mazzanti M, Su HT,et al. Automatic quantification of ejection fraction from gatedmyocardial perfusion SPECT. J Nucl Med 1995;36:2138-47.

11. Johnson LL, Verdesca SA, Aude WY, Xavier RC, Nott LT,Campanella MW, et al. Postischemic stunning can affect leftventricular ejection fraction and regional wall motion on post-stress gated sestamibi tomograms. J Am Coll Cardiol 1997;30:1641-8.

12. Hansen C. Digital image processing for clinicians, part II: filtering.J Nucl Cardiol 2002;9:429-37.

13. Hansen CL, Kramer M, Rastogi A. Lower accuracy of Tl-201SPECT in women is not improved by size-based normal databasesor Wiener filtering. J Nucl Cardiol 1999;6:177-82.

14. King MA, Glick SJ, Penney BC, Schwinger RB, Doherty PW.Interactive visual optimization of SPECT prereconstruction filter-ing. J Nucl Med 1987;28:1192-8.

15. Hansen C. Digital image processing for clinicians, part III: SPECTreconstruction. J Nucl Cardiol 2002;9:542-9.

16. Hansen CL. The role of the translation table in cardiac imagedisplay. J Nucl Cardiol 2006;13:571-5.

17. Eisner R, Churchwell A, Noever T, Nowak D, Cloninger K, DunnD, et al. Quantitative analysis of the tomographic thallium-201myocardial bullseye display: critical role of correcting for patient
motion. J Nucl Med 1988;29:91-7.
18. Klein JL, Garcia EV, DePuey EG, Campbell J, Taylor AT,Pettigrew RI, et al. Reversibility bull’s-eye: a new polar bull’s-eyemap to quantify reversibility of stress-induced SPECT thallium-201 myocardial perfusion defects. J Nucl Med 1990;31:1240-6.

19. Wackers FJ. Artifacts in planar and SPECT myocardial perfusionimaging. Am J Card Imaging 1992;6:42-58.

20. Bateman TM, Berman DS, Heller GV, Brown KA, Cerqueira MD,Verani MS, et al. American Society of Nuclear Cardiology positionstatement on electrocardiographic gating of myocardial perfusionSPECT scintigrams. J Nucl Cardiol 1999;6:470-1.

21. Cullom SJ, Case JA, Bateman TM. Electrocardiographically gatedmyocardial perfusion SPECT: technical principles and qualitycontrol considerations. J Nucl Cardiol 1998;5:418-25.

22. DePuey EG, Nichols K, Dobrinsky C. Left ventricular ejectionfraction assessed from gated technetium-99m-sestamibi SPECT.J Nucl Med 1993;34:1871-6.

23. Smanio PE, Watson DD, Segalla DL, Vinson EL, Smith WH,Beller GA. Value of gating of technetium-99m sestamibi single-photon emission computed tomographic imaging. J Am CollCardiol 1997;30:1687-92.

24. He ZX, Cwajg E, Preslar JS, Mahmarian JJ, Verani MS. Ejectionfraction determined by gated myocardial perfusion SPECT withTl-201 and Tc-99m sestamibi: comparison with first-pass radionu-clide angiography. J Nucl Cardiol 1999;4:412-7.

25. Hansen C. Digital image processing for clinicians, part I: basics ofimage formation. J Nucl Cardiol 2002;9:343-9.

26. Cooper JA, Neumann PH, McCandless BK. Effect of patientmotion on tomographic myocardial perfusion imaging. J Nucl Med1992;33:1566-71.

27. Friedman J, Berman DS, Van Train K, Garcia EV, Bietendorf J,Prigent F, et al. Patient motion in thallium-201 myocardial SPECTimaging. An easily identified frequent source of artifactual defect.Clin Nucl Med 1988;13:321-4.

28. Goerres GW, Burger C, Kamel E, Seifert B, Kaim AH, Buck A, etal. Respiration-induced attenuation artifact at PET/CT: technicalconsiderations. Radiology 2003;226:906-10.

29. Choi JY, Lee KH, Kim SJ, Kim SE, Kim BT, Lee SH, et al. Gatingprovides improved accuracy for differentiating artifacts from truelesions in equivocal fixed defects on technetium 99m tetrofosminperfusion SPECT. J Nucl Cardiol 1998;5:395-401.

30. Ficaro EP, Fessler JA, Shreve PD, Kritzman JN, Rose PA, CorbettJR. Simultaneous transmission/emission myocardial perfusion to-mography. Diagnostic accuracy of attenuation-corrected 99mTc-sestamibi single-photon emission computed tomography. Circula-tion 1996;93:463-73.

31. Fricke H, Fricke E, Weise R, Kammeier A, Lindner O, BurchertW. A method to remove artifacts in attenuation-corrected myocar-dial perfusion SPECT. Introduced by misalignment between emis-sion scan and CT-derived attenuation maps. J Nucl Med 2004;45:1619-25.

32. Grossman GB, Garcia EV, Bateman TM, Heller GV, Johnson LL,Folks RD, et al. Quantitative Tc-99m sestamibi attenuation-corrected SPECT: development and multicenter trial validation ofmyocardial perfusion stress gender-independent normal databasein obese population. J Nucl Cardiol 2004;11:263-772.

33. Weiss AT, Berman DS, Lew AS, Nielsen J, Potkin B, Swan HJ, etal. Transient ischemic dilation of the left ventricle on stressthallium-201 scintigraphy: a marker of severe and extensivecoronary artery disease. J Am Coll Cardiol 1987;9:752-9.

34. Hansen CL, Cen P, Sanchez B, Robinson R. Comparison ofpulmonary uptake with transient cavity dilation after dipyridamoleTl-201 perfusion imaging. J Nucl Cardiol 2002;9:47-51.

35. Hansen CL, Sangrigoli R, Nkadi E, Kramer M. Comparison of
pulmonary uptake with transient cavity dilation after exercise


thallium-201 perfusion imaging. J Am Coll Cardiol 1999;33:1323-7.

36. Chouraqui P, Rodrigues EA, Berman DS, Maddahi J. Significanceof dipyridamole-induced transient dilation of the left ventricleduring thallium-201 scintigraphy in suspected coronary arterydisease. Am J Cardiol 1990;66:689-94.

37. Gill JB, Ruddy TD, Newell JB, Finkelstein DM, Strauss HW,Boucher CA. Prognostic importance of thallium uptake by thelungs during exercise in coronary artery disease. N Engl J Med1987;317:1486-9.

38. Wackers FJT. On the bright side. J Nucl Cardiol 2005;12:378-80.

39. Berger HJ, Matthay RA, Loke J, Marshall RC, Gottschalk A, ZaretBL. Assessment of cardiac performance with quantitative radionu-clide angiocardiography: right ventricular ejection fraction withreference to findings in chronic obstructive pulmonary disease.Am J Cardiol 1978;41:897-905.

40. Brent BN, Mahler D, Matthay RA, Berger HJ, Zaret BL. Nonin-vasive diagnosis of pulmonary arterial hypertension in chronicobstructive pulmonary disease: right ventricular ejection fraction atrest. Am J Cardiol 1984;53:1349-53.

41. Williams KA, Schneider CM. Increased stress right ventricular
activity on dual isotope perfusion SPECT: a sign of multivessel
and/or left main coronary artery disease. J Am Coll Cardiol1999;34:420-7.

42. Reisman S, Maddahi J, Van Train K, Garcia E, Berman D.Quantitation of extent, depth, and severity of planar thalliumdefects in patients undergoing exercise thallium-201 scintigraphy.J Nucl Med 1986;27:1273-81.

43. Caldwell JH, Williams DL, Harp GD, Stratton JR, Ritchie JL.Quantitation of size of relative myocardial perfusion defect bysingle-photon emission computed tomography. Circulation 1984;70:1048-56.

44. Kaul S, Chesler DA, Okada RD, Boucher CA. Computer versusvisual analysis of exercise thallium-201 images: a critical appraisalin 325 patients with chest pain. Am Heart J 1987;114:1129-37.

45. Burow RD, Pond M, Schafer AW, Becker L. “Circumferentialprofiles:” a new method for computer analysis of thallium-201myocardial perfusion images. J Nucl Med 1979;20:771-7.

46. Weiss AT, Maddahi J, Lew AS, et al. Reverse redistribution ofthallium-201: a sign of nontransmural myocardial infarction withpatency of the infarct-related coronary artery. J Am Coll Cardiol1986;7:61-7.

47. DePace NL, Iskandrian AS, Hakki AH, Kane SA, Segal BL. Value ofleft ventricular ejection fraction during exercise in predicting the
extent of coronary artery disease. J Am Coll Cardiol 1983;1:1002-10.

myocardial perfusion and function

Documents