oxford medicine online principles of brain single-photon emission computed tomography ... · 2018....

22
Principles of brain single-photon emission computed tomography imaging Page 1 of 22 PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). © Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy ). Subscriber: Habib Zaidi; date: 27 November 2015 Publisher: Oxford University Press Print Publication Date: Oct 2015 Print ISBN-13: 9780199664092 Published online: Oct 2015 DOI: 10.1093/med/9780199664092.001.0001 Chapter: Principles of brain single-photon emission computed tomography imaging Author(s): Yong Du and Habib Zaidi DOI: 10.1093/med/9780199664092.003.0008 Oxford Medicine Online Oxford Textbook of Neuroimaging Edited by Massimo Filippi Principles of brain single-photon emission computed tomography imaging Introduction The historical development of medical imaging is marked by numerous significant technological accomplishments, driven by an unprecedented collaboration between multidisciplinary research groups. The first medical applications of tomographic imaging focused on the brain [1]. X-ray computed tomography (CT) and magnetic resonance imaging (MRI) have, for a long time, been the most widely-used imaging modalities for anatomic assessment of pathologic processes affecting the brain. Alternatively, radionuclide imaging methods, including single-photon emission computed tomography (SPECT) and positron emission tomography (PET) have emerged as useful medical imaging technologies for evaluating brain function. The development of the former technology for brain imaging dates back to 1976 when Drs Keyes and Jaszczak reported independently on the development of a brain SPECT system based on Anger camera mounted on a rotating gantry [2,3]. As nuclear medicine imaging has become integrated into clinical practice, several design trends have developed; with systems now available with a spectrum of features, from those designed for

Upload: others

Post on 22-Aug-2020

7 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 1 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

Publisher: OxfordUniversityPress PrintPublicationDate: Oct2015PrintISBN-13: 9780199664092 Publishedonline: Oct2015DOI: 10.1093/med/9780199664092.001.0001

Chapter: Principlesofbrainsingle-photonemissioncomputedtomographyimagingAuthor(s): YongDuandHabibZaidiDOI: 10.1093/med/9780199664092.003.0008

OxfordMedicineOnline

OxfordTextbookofNeuroimagingEditedbyMassimoFilippi

Principlesofbrainsingle-photonemissioncomputedtomographyimaging

IntroductionThehistoricaldevelopmentofmedicalimagingismarkedbynumeroussignificanttechnologicalaccomplishments,drivenbyanunprecedentedcollaborationbetweenmultidisciplinaryresearchgroups.Thefirstmedicalapplicationsoftomographicimagingfocusedonthebrain[1].X-raycomputedtomography(CT)andmagneticresonanceimaging(MRI)have,foralongtime,beenthemostwidely-usedimagingmodalitiesforanatomicassessmentofpathologicprocessesaffectingthebrain.Alternatively,radionuclideimagingmethods,includingsingle-photonemissioncomputedtomography(SPECT)andpositronemissiontomography(PET)haveemergedasusefulmedicalimagingtechnologiesforevaluatingbrainfunction.Thedevelopmentoftheformertechnologyforbrainimagingdatesbackto1976whenDrsKeyesandJaszczakreportedindependentlyonthedevelopmentofabrainSPECTsystembasedonAngercameramountedonarotatinggantry[2,3].Asnuclearmedicineimaginghasbecomeintegratedintoclinicalpractice,severaldesigntrendshavedeveloped;withsystemsnowavailablewithaspectrumoffeatures,fromthosedesignedfor

Page 2: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 2 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

clinicalwhole-bodyapplicationstoothersdesignedspecificallyforveryhigh-resolutionbrainresearchapplications[4].

SPECTisanon-invasive3Dtomographicimagingmodalitythatiswidelyusedinbothclinicaldiagnosisandacademicresearchtoassessmanydiseases[5,6,7].InSPECTimaging,the3Ddistributionofaradionuclide-labelledagentadministeredtoapatientismeasuredbydetectingthegamma-photonsemittedfromthedecayoftheradioactiveisotopesattachedtotheradiotracer.BecauseSPECTagentscanbedesignedtotargetspecificphysiologicalfunctions,SPECTimagescanprovideinformationonphysiologicalandphysiopathologicalprocessesatamolecularlevel.Thisuniquefeaturemakesitausefultoolforinvivoimagingofhumanbrainfunction,especiallyforstudyingdysfunctionoftheneurotransmissionsystemthatisrelatedtomanybraindiseases[8].Asaresult,brainSPECTimagingbecameapowerfultoolfordiagnosis,prognosis,evaluationofresponsetotherapy,andchoiceoftreatmentplanformanyneurodegenerativediseases[9,10].

Brainsingle-photonemissioncomputedtomographyinstrumentationConventionalSPECTimagingsystems

SincethepurposeofSPECTimagingistomeasureandrepresentthe3Ddistributionoftheradiopharmaceuticalsadministeredtothepatient,theenergiesofgammarayphotonsemittedbylabellingradionuclidesarehighenoughtoallowpenetrationthroughthepatient’sbody.Theemittedphotonscanbedetectedusingtheso-calledgammacamera(Fig.8.1)togenerateatwo-dimensional(2D)snapshotplanarimage.The2Dplanarimageisreferredtoasprojectionimagesinceitrepresentsaprojectedviewofthethree-dimensional(3D)radiotracerdistributionatacertainangle.Toacquirea3Ddistribution,aseriesofprojectionimagesaretakenaroundthepatientatdifferentdiscreteanglesbyrotatingthegammacameraaroundthepatientalongthesuperior-inferioraxis.Afteracquisition,the3Ddistributioncanthenbereconstructedfromthemeasuredprojectionimagesusingoneoftheavailablereconstructiontechniques.

Fig.8.1SchematicrepresentationofaSPECTgammacameraanditsmaincomponents.

Usually,conventionalSPECTimagingsystemshavetwoorthreegammacamerasthataremountedonagantrywhichprovidesmechanicalsupportandrotationofthecameras.Themaincomponentsofagammacameraincludethecollimator,scintillationcrystal,andphotomultipliertubes(PMTs)(Fig.8.1).

Thecollimatorisusedtoselectivelydetectphotonstravellinginacertaindirectionbecausethisinformationisneededforimagereconstruction.Itismadeofhighlydensephotonabsorbing

Page 3: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 3 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

material,usuallyleadortungsten,withapatternofholesthatallowsonlyphotonstravellingparalleltothehole-directiontopassandreachthedetectorcrystal.Theholescanbearrangedwiththesamedirection(parallelholecollimator)orconvergetocertainpoints(forexample,aconebeamcollimator).Thescintillatorcrystal,commonlysodiumiodide(NaI)crystal,willabsorbthephotonandtransferitsenergyintovisibleorultraviolet(UV)lightphotonsthroughaprocesscalledscintillation.ThenumberofvisibleorUVphotonsisproportionaltotheenergyofincidentphotons.Thevisible/UVphotonswilltraveltothePMTtroughalightguideandproduceanelectriccurrentsignal.ThePMTcanmultiplythecurrentsignalbyafactorofasmuchasamillion.Thesignalwillthenbefurtheramplifiedthroughtheamplifierandsenttoprocessingelectronicstogeneratethepositionandtheenergyinformationoftheincidentgammaphoton.Theinformationwillthenbepassedontoanacquisitionboard/computertogeneratetheprojectionimages.

ManyoftheclinicalSPECTsystemscommerciallyavailabletodayaredual-modalitySPECT/CTsystemsthatalsoincludeaCTsubsystem[11].TheSPECTimagesandCTimagesarecomplimentarytoeachother:SPECTimagesprovidefunctionalinformation,whiletheCTimagesprovideanatomicalinformationaboutlocationofdiseasesite.Byfusingthetwoimagestogether,anaccurateinterpretationofimagescanbeachieved.Inaddition,CTimagescanalsobeconvertedintoattenuationmapsrequiredforattenuationcompensationtoimproveSPECTimagequalityandquantitativeaccuracy.AplethoraofCTscannersareusedoncombinedSPECT/CTsystems.SomeusediagnosticqualityCTtoprovideimagessuitableforclinicalusage.Toreducepatientradiationdose,low-doseCTscannershavealsobeendeployed,mostlyfortypicalnuclearmedicineapplications[12].Theimagequalityoflow-doseCTisusuallylowandcanonlybeusedforattenuationcompensationandlocalization.However,withtheadvanceofiterativeCTreconstructionalgorithms,low-doseCTscannowalsoprovideimagesofgoodquality[13].

Dedicatedcollimatorsandhardwareforbrainimaging

MostclinicalSPECTsystemscanbeusedforvariousapplicationsthroughappropriateselectionofthemostsuitedcollimatorforaparticularstudy.ThecollimatoristhemostimportantpartofaSPECTcameraandisconsideredtobethemaincomponentaffectingspatialresolution,sensitivity,andimagenoise.Thechoiceofthecollimatordependsontheimagedobjectandtrade-offconsiderationsbetweensensitivityandspatialresolution.Usually,collimatorswithbetterspatialresolutionhavelowersensitivity,thusrequiringalongacquisitiontimeasotherwisetheimagewillbeverynoisy.Ontheotherhand,collimatorswithhighersensitivitywillprovideimageswithlowerresolution.Themostcommonlyusedcollimatorsareparallel-holecollimators.Forexample,inbrainimagingusing Tc-labelledcompounds,alow-energyhigh-resolution(LEHR)parallel-holecollimatorisoftenusedtoprovideimageswithhighspatialresolution.BecausehighspatialresolutionisdesirableinbrainSPECTimaging,muchworthwhileresearchfocusedondesigningdedicatedcollimatorstoimprovespatialresolutionwithoutsacrificingsensitivity[4].

Mostofthesecollimatorsareconvergingcollimators,inwhichtheholesarefocusedtoaline(fanbeamcollimator)orapoint(conebeamcollimator)infrontofthecollimator.Anumberofvaryingfocallengthcollimators,wheretheholesarefocusedtodifferentpointsinspacehavealsobeendevelopedandevaluatedinclinicalsetting.Thedetectionefficiencyofconvergingbeamcollimatorsishigherthanthatofparallel-holecollimators.Thespatialresolutionisalsoslightlybetter.Pinholecollimatorshavealsobeendesignedforimagingsmallorgans,suchasthethyroidortheextremitiestoprovideabetterspatialresolution,butattheexpenseofdetectionefficiency.Theefficiencycanbesubstantiallyimprovedbyusingmultiplepinholes[14].

99m

Page 4: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 4 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

Furthermore,eveninthepresenceofhighcountstatistics,themechanicalmotionsinvolvedandthebulkofthedetectorheadslimittheshortesttimeinwhichacompletesetofprojectionscanberecorded[15,16].NumerousattemptshavebeencarriedouttoconsiderablyboostbrainSPECTsensitivityusingseveraldetectorheadsorring-likearrangementofdetectorstoenabletheveryrapidacquisitionofafullsetofprojections.AlthoughtheuseofsuchdedicatedsystemsisnotprevalentintheclinicforbrainSPECTimaging,theyareusefulforresearchapplicationsandcouldalsobeusefulfortheassessmentofthepotentialroleofadvancesininstrumentationinthefuture[17,18].AgoodexampleoffullringdetectorsystemsistheFASTSPECT,developedattheUniversityofArizona,whichconsistsof24position-sensitiveNaI(Tl)detectorsthatarecompletelystationary,togetherwithastationarysetof2-mmpinholecollimators,hence,achievinghighsensitivity(fastdynamicbrainscans)andhighspatialresolution[19].Theringdetectorsfromstationarysystemssurroundthehead,whichmakesitpossibletoacquiredatafrom360°atthesametime.Thisisespeciallyattractivefordynamicimagingwhereashortrepeat-scanoveraperiodoftimeisperformed.AnotherdedicatedbrainscannerwithstationaryannularNaI(Tl)detectoristheCERASPECTsystem,developedbyDigitalScintigraphicsInc.[20],whichisequippedwitharotatingcollimator.AmodifiedversionofthiscollimatoristheSensOgrade,avariablefocusingcollimator,whichsamplestheprojectionsunequally,withcentralregionsmoreheavilyrepresentedtocompensateforattenuationfromcentralbrainstructures,thusyieldingafour-tonine-foldincreaseinsensitivitycomparedwithconventionaldual-headcameras[21].AthirdexampleistheSPRINTIIsystemdevelopedattheUniversityofMichigan,whichconsistsof11detectorsarrangedinapolygonalfashionandarotatingcollimator,whichallowstheacquisitionofacompletesetoffan-beamprojectiondatawithin1/12ofarotation[22].AnotherexampleofunconventionalsystemsistheNeuroFocus™multi-conebeamimager(NeurophysicsCorporation,Shirley,MA,USA),whichproducestomographicSPECTimageswithanintrinsicspatialresolutionof~3mm.TheoperationoftheNeuroFocus™high-definitionfocusingemissiontomographicscanner(HDFET)followsthesameprinciplesofscanningopticalmicroscopestoobtainhigh-resolution3Dimagesofbiologicaltissue[23].Ahighlyfocusedpointoflightisscannedmechanicallyinthree-dimensionstouniformlysamplethevolumeunderobservation.AsanalternativetodynamicSPECTimagingusingmultiple,fastrotations,strategiesinvolvingtheuseofonlyasingle,slowcamerarotationhavebeenproposed[24,25].Furtheradvancesinelectronicsarepermittingnewcountingstrategiesandadvancesinelectroniccomponentcapabilityareallowingforenhancedsensitivity[26].

DataacquisitionprotocolsDataacquisition

DuringSPECTimageacquisition,aseriesof2Dprojectionimagesaretakenatdifferentanglesbyrotatingthecameraaroundthepatient.Toacquireenoughdataforimagereconstruction,itisimportanttoselectthecorrectscanarc,whichdefinestherangeofrotationandwheretostart.Forexample,forcardiacimaginga180°rotationfromrightposteriorobliquetoleftanteriorobliqueisused,whileforbrainimaginga360°rotationisusuallyrequired.Onealsoneedstodecidetheorbittospecifythedistanceofthecamerafromthepatientateachangle.Theorbitcanbeeithercircularornon-circular.Non-circularmeansthatthedistancesofthecamerafromthecentreofrotation,termedastheradiusofrotation,arenotthesamefordifferentangles.Inacircularorbit,theradiusofrotationisconstant.Foraparallel-holecollimator,thecloserthepatienttothecamera,thebetterthespatialresolutionis.Therefore,anon-circularorbitisoftenusedforimagingthetorso,wherethecameraispositionedasclosetothepatientaspossible.InmodernSPECT

Page 5: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 5 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

systems,thiscanbedoneautomaticallyusingopticalsensorsinstalledonthecamera.Forbrainimaging,duetotheroundshapeofthehead,acircularorbitisoftenusedwheretheoperatorcanspecifytheradiusofrotation.

Thegammaphotonsemittedfromradioactivedecayhavecharacteristicenergy/energies,calledpeaksthatarespecifictotheisotope.Forexample, Tcdecayemitsgammaphotonsat140.5keVonly,while Idecayemitsgammaradiationmainlyat159keVwithsomeotherhigherenergypeakswithlowabundance.Ideally,duringSPECTimagingonlyphotonsdetectedwiththeexactpeakenergywillbecountedastrueevents.However,duetothelimitedenergyresolutionofthegammacamera,theenergiesofdetectedphotonsareGaussiandistributedaroundthepeakenergy.Tomaintainareasonablesensitivity,photonsdetectedwithenergiesinacertainrangearoundthepeakenergywillallbecountedastrueevents.Thisenergyrangeiscalledtheacquisitionenergywindow.Ontheotherhand,photonsscatteredinthepatientbodywillcontaminatetheimagesandtheyusuallyhaveabroadenergydistributionlowerthanthepeak.Toreducethenumberofrecordedscatteredphotons,theenergywindowshouldnotbetoowide.Thechoiceoftheenergywindowisdictatedbyabalancebetweenrecordingasmanypeakphotonsaspossiblewithoutsignificantlyincreasingscatteredphotoncounts.Commonly-usedenergieswindowsareusually15–20%wideandcentredaroundthepeakenergy.

AnotherimportantparameterinSPECTistheacquisitiontimeforeachprojectionangle.Longacquisitiontimesresultinhighcountsandlowimagenoise.However,thiswillalsoreducepatientcomfortandincreasechancesofartefactscausedbypatientmovementduringacquisition.Overall,thechoiceofscanningparametersdependsontheimagingtaskathand.Theaimistoacquireasmuchcompletedataaspossibletoprovidegoodimagequalityandmaketheprocesscomfortabletopatients.MostcommercialSPECTsystemshavemanufacturerpresetorrecommendedacquisitionprotocolsforcommonapplications,suchascardiacimaging,brainimaging,tumourimaging,andbonescans.Newprotocolscanalsobesetbytheuserstoaccomplishtheirspecialneeds.Thisrequirestheavailabilityofaqualifiedmedicalphysicisthavingtherequiredskills,expertise,andknowledgeofthistechnology.

Dual-isotopeimaging

Insomesituations,itmaybenecessarytoacquireimagesofmorethanonephysiologicalfunctiontoprovideaccuratediagnosis.Forexample,inbrainimagingofthedopaminergicsystem,itisgenerallyrecognizedthattheanalysisoftheintegrityofboththepre-andthepost-synapticneuroniscrucialindistinguishingdifferentparkinsoniansyndromes.Thepresynapticneuronscanbeimagedwithanagenttargetingthedopaminetransporter(DAT)onthecellmembrane.Agentstargetingthedopaminereceptors,especiallytypeIIreceptor(D2R),couldbeusedinimagingthepost-synapticneurons[27,28].Traditionally,DATandD2RimagingareperformedondifferentdaysbecausemostofDATandD2Ragentsarelabelledwiththesameisotope( I).Sincethehalf-lifeof Idecayis13hours,thetwoscansmustbeseparatedbyseveraldaystoallowtheradioactivityfromthefirstscantodecayinordertoavoidcontaminationofthesecondscan.Duringthisinterval,thepatient’sphysiologicalstatusmayhavechanged.Moreover,thepatient’spositionduringthetwoscanswillprobablybedifferent,whichwillcauseregistrationerrorsbetweentheimages[29].Thisisparticularlyrelevantforpatientssufferingtremors,wheretheartefactsinthetwoimagesmayhavedifferentpatternsthatwillfurthercomplicatediagnosis.

Thesedrawbackscan,however,beovercomebyusingsimultaneousdual-isotopeacquisition

99m123

123123

Page 6: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 6 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

methods.Takingadvantageofmoderncameras’higherenergyresolutionandtheabilitytoacquiremultipleenergywindows,simultaneousdual-isotopeimagingisamethodthatcanbeusedtodiagnosediseasesbyimagingdifferentphysiologicalinformationrevealedbyagentslabelledwithdifferentisotopesthatemitgammaphotonsatdifferentenergies.Comparedtoseparateimaging,simultaneousacquisitionreducesacquisitiontime,patientdiscomfortandmotionartefacts.Perhaps,moreimportantly,simultaneousacquisitionensuresperfectregistrationoftheimagesfrombothisotopesintimeandspace.Furthermore,motionartefacts,ifany,willbethesameonimagesofbothisotopes.Forexample,withtheintroductionof Tc-labelledDATligands,suchas Tc-TRODAT,SPECTimagingoftheDATandD2Rcannowbecarriedoutsimultaneouslyusingdual-isotope Tc/ Iimaging.Thereisalsointerestinsimultaneousimagingofbrainperfusionusing

Tc-HMPAOandneurotransmissionwith I-IBZM.However,researchonsimultaneousTc/ Idual-isotopebrainimaginghasshownthatimagequalityissignificantlydegradedbythe

cross-talkcontaminationbetweenthetwoisotopes.Cross-talkherereferstothecontaminationofoneagent/isotope’simagecausedbytheotherisotopeusedinsimultaneousdual-isotopeimaging.Beforesimultaneousimagingcanbeadoptedclinically,anefficientcross-talkcompensationmethodmustbedevelopedandvalidated[30,31,32].

ImagereconstructionImagereconstructiontechniques

SPECTimageformationcanbeexpressedasanintegrationofthe3Dradioactivitydistributionalongeachprojectiondirection,i.e.lineofintegration,intoaseriesof2Ddata.Imagereconstructionisaninverseproblemthattriestoestimatethe3Dactivitydistributionfromthe2Dprojectiondata.Bothanalyticalanditerativereconstructionapproacheshavebeendevisedtosolvethisreconstructionprobleminordertoprovidethebestestimatesthatareasclosetothetruthaspossible.Thesetwocategoriesofreconstructionstrategiesarebrieflysummarized.

Analyticalreconstruction

Asdescribedabove,aprojectionisanintegrationoftheactivitydistributionalongthelineofintegration.ThecollectionoftheseprojectionsasafunctionofprojectionangleisreferredtoastheRadontransformoftheobject[33].Byinversingtheradontransformoftheprojections,theoriginal3Dactivitydistributioncanthenbereconstructedanalytically.Themostwidelyusedanalyticalreconstructionmethodisthefilteredback-projection(FBP)algorithm,whichisstillusedinmanyimagereconstructiontasks.

Analyticalreconstructionmethods,suchasFBP,focusonthegeometryandsimplifythephysics.Theyaresimple,fast,andcanprovideaccurateresultswhentheassumedprojectionmodelmatchestheimageformationprocess.However,theyaresensitivetonoiseandmissingdata,whichwillresultinartefactsandlossofreconstructedimageresolution.Anothermajordisadvantageofanalyticalmethodsisthattheydonotallowprecisemodellingofthephysicalandstatisticalcharacteristicsofthedataacquisitionprocess,thusresultinginimageartefactsandpoorquantitation.Thisproblemcan,however,besolvedthroughtheuseofiterativereconstructionalgorithms,whereimagedegradingfactorscanbemodelledduringthereconstructionasdescribedinthenextsection.

Iterativereconstruction

99m 99m

99m 12399m 12399m 123

Page 7: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 7 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

Fig.8.2showsageneralizedflowchartofaniterativereconstructionalgorithm.Aniterativemethodcanberegardedasanoperatorworkingbetweentheimagespace,i.e.the3Dspacerepresentingtheactivitydistributionintheobject,andprojectionspace,whichrepresentstheSPECTmeasurementofthe2Dprojections.Inaniterativemethod,aninitialestimateisprojectedtogenerateanestimatedprojectionofthecurrentestimate.Estimatedprojectionsarethencomparedwiththemeasuredprojections.Theerrorbetweenthetwoisback-projectedtotheimagespace,thenusedtoupdatethecurrentestimate.Byrepeatingtheprojection–back-projectionprocessinafeedbacklooptoupdatetheimageestimateuntilagivencriterionisfulfilled,iterativemethodscanproducehigh-qualityimageswithimprovedresolutionandnoiseproperties,andresultinadequatequantitativeaccuracy.

Fig.8.2Flowchartofaniterativereconstructionalgorithm.

Amongalliterativereconstructionalgorithms,themostsuccessfulandpopularonesarestatisticalreconstructions,whichareoftensimplyreferredtoasiterativereconstructionapproaches.Aniterativestatisticalreconstructionmethodconsistsofthreemajorcomponents:

1.Anunderlyingstatisticalmodelwithassociatedobjectivefunction.2.Aniterativealgorithmtofindtheoptimalestimateintermsoftheobjectivefunction.3.Amodeloftheimageformationprocessoftenimplementedusingaprojector–back-projectorpair.

Amongthese,theprojector–back-projectorpairisveryimportant.Itservesasabridgeconnectingtheimagespace(estimate)andtheprojectionspace(measurement).Itisintheprojection–back-projectionoperatorsthattheexactmodellingofthephysicsandimageformationprocessisperformed.Throughtheiterativeprocess,themodellingallowsforthecompensationofimage-degradingfactors,suchasscatterandattenuation.Theunderlyingstatisticalmodelandresultingobjectivefunctionplaytheroleofdecidinghowwelltheestimatematchestheprojections,takingintoaccountthenoisepropertiesofthemeasureddataandknowledgeaboutthecharacteristicsoftheunderlyingactivitydistribution.

ThemostpopulariterativereconstructionalgorithminSPECTisthemaximum-likelihoodexpectation-maximization(ML-EM)algorithmanditsacceleratedvariation,theordered-subsetsexpectation-maximization(OS-EM)approach[34,35].TheML-EMalgorithmisbasedonthefactthatphotonemissionanddetectionfromradioactivedecayarePoissondistributed.Itattemptstomaximizethestatisticallikelihoodthatthemeasuredprojectionscamefromthereconstructed

Page 8: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 8 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

image.InOS-EM,thealgorithmismadefasterbyusingonlypartoftheprojectiondata,calledasubset,duringeachupdate.ComparedwithML-EM,OS-EMconvergesfasterwithafactorapproximatelyequaltothenumberofsubsetsused.ThiscategoryofreconstructionstrategieswassosuccessfulthatmostcurrentSPECTsystemsoffersoftwareforenduserstoperformML-EMandOS-EMreconstructionwithvariousphysicsmodellingoptions.Whilethishasbeenshowntobeofgreatpracticalvalue,itshouldbenotedthatOS-EMhasarelativelyweaktheoreticalbasisanditsconvergencepropertiesareonlypoorlyunderstood.

Despitetheiradvantagesandattractiveproperties,thesetechnicaladvanceshavemanylimitations.Iterativemethodscanbeverycomputationallydemanding,particularlywhencomplexmodelsareusedtomodelthephysicsoftheimageformationprocess.However,thewidespreadavailabilityofhighperformancecomputing,evenondesktopcomputers(includinggraphicsprocessingunitandcloudcomputing),andthecontinuingdevelopmentofrapidlyconvergingalgorithmshasledtorenewedinterestiniterativetechniquesandmadethemahottopicforleadingmanufacturersandacademicresearchgroups.

Imagedegradingfactors

Ideally,onlyphotonswithanoriginalpaththatisalongthecollimatorholedirectioncanbedetectedintheprojections.However,whengamma-photonstravelinsideanobject,theyinteractwithmatterthroughanumberofphysicalprocesses,suchasphotoelectricabsorptionandComptonscatter[36].Theprobabilityofoccurrenceofeachprocessdependsontheenergyofthephotonandtheelectrondensityofthematerial.Thoseinteractionscouldreducethenumberofphotonsorchangeaphoton’spathandenergy.Thedetectionofthosealternatedphotonscancausefalseinformationanddegradetheimagequalityinmultipleways.Fig.8.3demonstratesthedifferentfactorsthatcandegradeSPECTimages.Amongthose,themostsignificantfactorsarephotonattenuation,scatter,andcollimator-detectorresponse.

Fig.8.3OverviewofphotondetectioninSPECTrepresentingdifferentphysicaldegradingfactors.(1and2)Photonspassthroughthecollimatorwithoutinteractions,1alsoshowstheattenuationofthephotonintensity.(3)Photonpenetratesorscattersinthecollimator-detectorsystem.(4)Photonscattersinsidetheimagingobject.(5)Photoelectricabsorption;and(6)Photonescapestheobjectwithouthittingthecollimator-detectorsystem.

Whenphotonspassthroughtheimagingobject,theycanbeabsorbedbyphotoelectricabsorption

Page 9: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 9 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

ordivertedbyComptonscatter.Asaresult,theintensityofthephotonfluxfinallyreachingthedetectorwillbemuchsmallercomparedwiththeoriginaloneandthiseffectisreferredtoasphotonattenuation.Attenuationreducesthemeasuredprojectioncounts.Thedegreedependsonthephotonpathandtheobject.Itcanbeexpressedasanexponentfunctionoftheproductoftheattenuationcoefficientoftheobjectandthedistancethephotonhastotravelintheobject.Ifnotcompensatedfor,attenuationleadstoreconstructedimageswithreducedintensityinvoxelsdeepintheobject.Itcanalsoresultinvisibleartefactsandquantificationwithsignificanterrors.

Anumberofphotonsemittedfromtheradionuclidewillscatterinthepatientatleastoncebeforeexitingthebody.ThedominantscattereventsareComptonscatter,whichreflectsacollisioninteractionbetweenagamma-photonandanoutershellelectronofanatom.Duringthecollision,gamma-photonswilltransfersomeoftheirenergytotheelectronandchangethedirectionofmovement.Theenergyofthescatteredphotonisafunctionofthescatteringangle,givenbytheComptonformula.Thephotonlosesmostofitsenergywhenitisscatteredbackward(180°).Whenthescatteringangledecreases,theenergyofthescatteredphotoncomesclosetotheincomingphoton’senergy.TheprobabilityofComptonscatteringdependsonthephotonenergyandthenumberofelectronsavailableintheobject.Theprobabilityoftheangulardistributionofthescatteredphoton,describedbyKlein–Nishinaformula,isalsoafunctionofthephotonenergyandscatteringangle[36].

BecauseComptonscatteringreducesphotonenergy,usuallyonlythosephotonsscatteredonceortwice,andwithinasmallscatteringanglecanbedetectedwithintheimagingenergywindow.Sincescatterdivertsthedirectionofscatteredphotons,whenthosephotonsaredetected,theyresultincountsinvoxelsthattheywouldnotreachwithoutbeingscattered.Scatterinprojectiondataisspatiallyvaryingandisafunctionoftheobjectattenuationproperties,theimagingenergywindow,andthesourcedistribution.Inbrainimagingusing I-or Tc-labelledcompounds,scatteredphotonscancontributeupto11%ofthetotalcountsinthefinalimages.Ifnotcompensatedfor,scattercouldresultinastructuredbackgroundartefactthatresemblesthemedium-andlow-frequencycharacteroftheactivitydistribution,andassuchreducesimagecontrast.Quantitatively,scattercanleadtooverestimationoftraceruptake[37,38].

Thecollimatoristhecrucialelementdeterminingthesensitivity,spatialresolutionandcontrastofSPECTimages.Aperfectcollimatoronlyallowsthedetectionofphotonsthattravelinadirectionparalleltothecollimatorhole.AsshowninFigs8.3and8.4,sincearealcollimatorholehasafinitegeometricsize,photonsincidentwithinasmallacceptanceanglefromthehole-directioncanstillbedetected.Thisisreferredtoasgeometrically-collimatedphotonsorcollimatorgeometricresponse.Inaddition,duetostatisticvariations,thepositionofphotonabsorptioninthecrystalisusuallydeterminedwithimprecision,andisdefinedasintrinsicresolution.Bothcollimatorgeometricresponseandintrinsicresolutionreducethespatialresolutionandcauseblurringoftheimage.Theseeffectscanbemodelledusinggeometricresponsefunctions.Typically,Gaussianfunctionsareusedtoanalyticallymodelthegeometricresponses.ThefullwidthathalfmaximumoftheGaussianis,therefore,oftenusedasanindicationofthesystemspatialresolution[39].Ingeneral,collimatorswithlargeholesallowmorephotonstopassthrough,butwillalsocausemorespatialblurring.Collimatorswithsmallholescanprovidebetterspatialresolution,butlowersensitivity.Theresolutionisalsoafunctionofdistance:thefarthertheobjectfromthecollimator,theworsetheresolution.

123 99m

Page 10: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 10 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

Fig.8.4Differentcomponentsofthecollimator-detectorresponse.(1)Geometricresponse.(2)Collimatorpenetration.(3)Collimatorscatter.

Photonsincidentonthecollimatorwithlargeanglesareusuallyabsorbedbytheseptabetweencollimatorholes.However,somehigh-energyphotonsemittedfromisotopessuchas IcanstillpenetratetheseptaorpassthroughafterscatteringinsidetheseptaasshowninFig.8.4.Thepenetratedphotonsfromapointsourceusuallyresultinastar-orstripe-shapedpatternintheprojection.Collimatorscatteredphotonsusuallyresultinabroadlydistributedbackgroundintheprojection.Ifnotcompensatedfor,thesephotonscancauseartefactsinthereconstructedimages.Thefullcollimator-detectorresponse,includinggeometricresponse,collimatorpenetration,andcollimatorscatterdependsonthecollimatorgeometryandthephotonenergy.Itcanbemodelledbythecollimator-detectorresponsefunction(CDRF)thatismeasuredorsimulatedbyplacingapointsourceatvariousdistancesfromthefaceofthecollimator.

Fig.8.5showssimulatedSPECTimagesconsideringthedifferentimagedegradingfactorsdiscussedabove.TheimageswerereconstructedusingFBPwithoutpost-reconstructionfiltering.Fromlefttoright,theimagequalityisgettingworsebecausemoredegradingfactorswereincludedinthesimulationprocess.Theworstimageisthelastoneontheright,whichincludesallthephysicaldegradingfactorsandisthemostrepresentativeofwhatisfoundintheclinicwithoutanycompensation.Theimagesalsoindicatethatattenuationreducestheintensityatthecentreofthebrain,collimator-detectorresponseblursimagesandreducesspatialresolution,andscatterreducesimagecontrast.

Fig.8.5

123

Page 11: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 11 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

Effectsofimagedegradingfactors.Theimagesrepresenttheactualtracerdistributioninthestriatalbrainphantom(a)andreconstructedSPECTimagesgeneratedwithoutdegradingfactors(b);withphotonattenuation(c);withattenuationandcollimator-detectorresponseblurring(d);andwithattenuation,collimator-detectorresponseblurring,andscatter(e).

ThespatialresolutionintypicalSPECTimagesisaround1–2cm.AsshowninFig.8.5,thefinitespatialresolutionresultsinblurredimageswithpartialvolumeeffects(PVEs).PVEsincludebothalossofcountsinstructuressmallerthantwoorthreetimesthesystemresolution’sfullwidthathalfmaximum(partialvolume)andcontaminationbetweenadjacentregions(spillover).PVEsalsodependontheshape,size,andrelativeactivitiesoftheobjectsofinterest.Theproceduresappliedafterimageacquisition,suchasimagereconstruction,willalsoaffectthelevelofPVEs.Inbrainimaging,duetothefinedimensionsofthestructuresofinterest,suchastheputamenandcaudate,PVEscancauselargequantitativeerrors.

Inaddition,SPECTimagesalsosufferfromnoise.NoiseinSPECTimagesconsistsmainlyofstatisticalnoiseresultingfromtherandomnatureofphotonemission(followingradioactivedecay)anddetectionprocesses.ItcanbecharacterizedbyaPoissondistributioninwhichthevarianceequalsthemean.Therefore,noiseinSPECTdataisafunctionofthenumberofdetectedcounts—thehigherthecounts,thelesserthenoise.Noisecouldsignificantlyaffectimagequalityandquantitativeaccuracy,andreducethedetectabilityofsubtleabnormalitiesandtheirmeasurement.Noisecanalsocauseimageartefacts,introducingtexturesintouniformly-distributedregions.Increasingthecountlevelusinglongacquisitiontimesorhighinjectiondosescanreducenoise.Lowpassfilteringisoftenusedtosmoothoutnoiseinthereconstructedimages.Alternatively,constrainscanalsobeappliedduringiterativereconstructiontoreducenoise.

Cross-talkindual-isotopeimaging

Insimultaneousdual-isotopeimaging,aphotonemittedfromoneisotopecanbedetectedinanotherisotope’senergywindowowingtothefiniteenergyresolutionofthedetectorsystem,andtheinteractionofphotonswiththeobjectandcollimator-detectorsystem.Thesephotonscanreduceimagequality,degradequantitativeaccuracy,andcauseartefactsintheimages.Thiseffectisreferredtoascross-talkcontamination.Cross-talkcontaminationdependsontheemissionenergiesandtheimagingenergywindowsofbothisotopes.Assuch,theoverallimpactisdifferentforvariousisotopecombinations,makingitdifficulttofindauniversalcompensationmethod.

Fig.8.6showssimulatedenergyspectraofasimultaneous Tc/ Idual-isotopebrainSPECTstudy.Thespectraindicatethatcross-talkfrom Tcinto Iprojectiondataoriginatesmainlyfromunscattered Tcphotonsthataredetectedinthe Ienergywindowasaresultofthedetector’sfiniteenergyresolution.Thecross-talkfrom Iinto Tcwindow,however,iscomplexandincludesphotonsoriginatingfromavarietyofprocesses.Inadditiontothe159keVphotonsusedforimaging, Idecayalsoemitshigh-energyphotonswithenergiesrangingfrom182keVto783keVwithatotalabundanceof~3%[40].Thedown-scatterofthesephotonsintoboth TcandIenergywindowscanfurthercontaminatetheimagesofbothradionuclides.

99m 12399m 123

99m 123123 99m

12399m

123

Page 12: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 12 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

Fig.8.6Simulatedenergyspectraofa Tc/ Idual-isotopestudy.

Compensationtechniques

Toimproveimagequalityandquantitativeaccuracy,imagingdegradingfactorshavetobecompensatedfor.Tocompensateforattenuation,anattenuationmaprepresentingthespatialdistributionofthepatient’sattenuationcoefficientsisrequired.TheattenuationmapcanbederivedbyscanningthepatientusingeithertransmissionsourcesofradionuclideshavinganenergyclosetotheenergyoftheradionuclideusedforSPECTimagingorfromCTimagesacquiredoncombinedSPECT-CTsystems.Inthepast,analyticalmethodshavebeenusedtocompensateforphotonattenuation,whichusuallyinvolvesscalingtheprojectionimageorpost-reconstructionfiltering[41,42].Becausethesemethodsoftenassumeuniformattenuationofthemedium,theycanonlyprovideanapproximatecompensation,andsometimesmayevencausemoreartefactswhentheobjectisnon-homogeneous[43].

Accurateattenuationcompensationcanbeachievedusingiterativereconstruction-basedcompensation,wheretheattenuationmapacquiredfromatransmissionscanisusedtomodeltheattenuationintheprojection–back-projectionoperators[44].Iterativereconstruction-basedattenuationcompensationhasbeenshowntosignificantlyimprovetheimagequalityandquantitativeaccuracy,andiscurrentlywidelyusedintheclinic.Similarly,anaccuratecompensationofthecollimator-detectorresponseisusuallyachievedthroughiterativereconstruction-basedcompensationbyincludingtheCDRFmodelsintheprojection–back-projectionoperators[45].

Comptonscattercompensationstrategiescanbecategorizedintomethodsthatcompensateforscatterdirectlyontheprojectiondata(pre-reconstruction),methodsthatcompensateforscatterduringthereconstructionprocessandpost-reconstructionrestorationfiltering[38].Methodsthatoperateonprojectiondatausuallyaccomplishthecompensationbysubtractingthescatterestimatefrommeasuredprojectiondata.Scatterisoftenestimatedbyscalingprojectiondataacquiredinoneortwoscatterwindowswithpre-determinedfactors[46,47,48].Thescattercanalsobeestimatedbyanalysingthespectrumofprojectiondataacquiredinmultiplenarrow(1–4keVwide)energywindowsusingspectralfittingtechniquesorartificialneuralnetworks[49,50].Post-reconstructionfilteringusuallycompensatesforthescatterbydeconvolvingthereconstructedimageswithascatterresponsefunction(ScRF)acquiredthroughexperimentalmeasurementsorderivedfromMonteCarlosimulations[51].Bothpre-andpost-reconstructionscattercompensationmethods

99m 123

Page 13: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 13 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

makesimpleassumptionsaboutthescattercomponent,suchasshift-invariance,whichresultsinbiasedscatterestimates.Furthermore,scattercompensationisachievedbysubtractionoftheestimatedscattercomponent,whichisusuallyaccompaniedbyalargeincreaseinstatisticalnoise[52,53].

Reconstruction-basedscattercompensation(RBSC)methodscompensateforscatterbymodellingthespatiallyvaryingandobject-dependentSRFintheprojection–back-projectionoperatorswithintheframeworkofaniterativereconstructionalgorithm[37,54].RBSCmethodswereshowntoprovidethemostaccuratescattercompensationamongallmethods[37,38].CurrentlyavailablescattermodellingtechniquesforRBSCincludetheslab-derivedscatterestimation(SDSE)model,theeffectivesourcescatterestimation(ESSE)techniques[54,55],thenon-stochasticnumericalintegrationmethod[56,57],andthefastMonteCarlosimulationapproach[58,59].

Todemonstratetheefficacyofthevariousimagecorrectiontechniques,Fig.8.7showsrepresentativeSPECTbrainimagesreconstructedwithcompensationforvariousdegradingfactors.Theimagequalitygraduallyimproveswhenmoredegradingfactorsarecompensatedfor.Thebestresultisachievedbyincludingcompensationsforallthephysicaldegradingfactors.

Fig.8.7Imagesreconstructedwithcompensationfordifferentphysicaldegradingfactors.Theimagesrepresenttheactualtracerdistributioninthestriatalbrainphantom(a)andreconstructedSPECTimagesobtainedwithoutcompensation(b);withattenuationcompensation(c);withattenuationandcollimator-detectorresponsecompensation(d);andwithattenuation,collimator-detectorresponse,andscattercompensation(e).

Mostofcross-talkcompensationmethodshavebeendevelopedbasedonscattercompensationmethods[60,61,62,63,64,65].Theproceduresincludeestimatingthecross-talk,andsubtractingitpriortoreconstructionorusingreconstruction-basedcompensationapproaches.Thecross-talkcanbeestimatedusingmultipleenergywindowmethods,includingthetriple-energywindowtechniqueormorecomplicatedtechniques,suchasthoseusingartificialneuralnetworks,constrainedspectraldeconvolution,andprincipalcomponentanalysis.Cross-talkmodelswerealsodevelopedforiterativereconstruction-basedcompensationthatuseestimatesoftheactivitydistributionforeachisotopecombinedwiththephysicsoftheimage-formationprocesstoestimatethecross-talk.Forexample,model-basedcross-talkcompensation(MBCC)usestheESSEtechniquetomodelphotoninteractionsinsideobjects,andcollimator-detectorresponsefunctionstomodelphotoninteractionswithinthecollimator-detectorandthefiniteenergyresolutionofthedetector[66].MonteCarlosimulationshavealsobeenusedforcross-talkestimation[58].

Fig.8.8showssimultaneous Tc/ Idual-isotopeimagesreconstructedwithoutcross-talkcompensationandwithMBCC,comparedwithcross-talkfreesingle-isotopeimages.Overall,withoutcompensation,thecross-talkreducedimagecontrastforboth Tcand Iimages.AftercompensationusingMBCC,theimagesareveryclosetosingle-isotopeimages,indicatingthe

99m 123

99m 123

Page 14: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 14 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

efficacyofcross-talkcompensation[66].

Fig.8.8Simultaneousdual-isotopeimages.Toprow: Tcimages,bottomrow: Iimages.Dual-isotopeimageswithoutcross-talkcompensation(aandd);withMBCC(bande);andcross-talkfreesingle-isotopeimages(candf).

Compensationforcollimator-detectorresponsecanimproveimageresolutionandreducePVEs.However,completecountrecoveryhasnotbeenachievedyet[45,66,67,68].Additionalpartialvolumeeffectcompensationisusuallyrequired.Aplethoraofpartialvolumecompensation(PVC)strategieshavebeenproposed[69,70].SomeofthesemethodsattempttoremoveorreducethePVEsbydeconvolvingtheimagewiththesystempointspreadfunction.Toimprovenoiseproperties,regularizationisrequiredduringdeconvolution.Pixel-by-pixeltemplate-basedapproaches,wherespill-inandspill-outcountsaremodelledbytheireffectontemplateimagesfollowedbycompensationusingsubtractionanddivision,werealsoproposed[71].AlternativeapproachesdirectlycompensateforPVEsattheregionalleveltoprovidecorrectedactivityestimatesusingatransfermatrixofPVEs[72].ApplicationofPVCtodopaminergicneurotransmissionSPECTimaginghasbeenshowntosignificantlyimprovequantitativeaccuracy[73].Reconstruction-basedPVCmethodshavealsobeenproposed,usingforinstance,themaximumaposterioriapproach[74].MethodsthatincorporatePVCdirectlyintothekineticmodellingprocessbyintroducingadditionalparametersthatmodelPVEswerealsoreportedfordynamicimaging[75].Overall,accuratePVCrequiresperfectknowledgeofthesystemresolutionandprecisedelineationofregions-of-interest(ROIs)boundaries,whichcanbeperformedthroughtheuseofregisteredhigh-resolutionanatomicalimagessuchasCTorMRI.

ImagequalityassessmentanddataanalysisQualitativeversusquantitative

Traditionally,SPECTimageshavebeenqualitativelyassessedbyvisualobservationoftraceruptake

99m 123

Page 15: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 15 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

inregionslinkedtoknownpatternsofvariousdiseases.Thevisualassessmentconveysinformationonwhetherthetraceruptakeisnormalorabnormal,and,ifabnormal,itprovidesanideaaboutthemagnitudeoftheabnormality.InbrainSPECT,thevisualassessmentprovidesinformationaboutrighttoleftasymmetriesandwhichstructuresarethemostaffectedbydisease.

However,becausetheuptakeofanimagingtracerisoftenstronglycorrelatedtotheintegrityofphysiologicalfunctionsintargetedROIs,thequantitativemeasurementsoftraceruptakeinthoseROIscanprovidemoreinformationforin-depthassessment.Mostimportantly,quantitativeanalysisenablestoestimatephysiologicalparameters,suchasbloodflow,metabolism,andreceptorconcentration.Thesemeasurementscanbeusedinclinicaldiagnosticwork-uporintheassessmentoftheefficacyoftherapies.Forinstance,thebindingofDATorD2Ragentsinthedopaminergicsystemarestronglycorrelatedtothedegreeofdiseaseprogressioninmovementdisorders.Recentresearchhasdemonstratedthatthequantitativemeasurementoftraceruptakecanprovidemoreinformationforbetterassessmentofmanybraindiseaseprocesses[10,76,77].

Quantificationtechniques

QuantitativestudiesusuallyinvolvedefiningROIsontheimagesandthenmeasuringthetraceuptakeinsidethem.OnecommonlyusedmethodinbrainSPECTistocalculatethespecificbindingpotential(SBP,alsocalledspecificuptakevalue)oftheradiotracerintheROIastheratiobetweenactivityconcentrationinsidetheROIandtheactivityconcentrationsinareferenceregionthathasnospecificbindingforthetracerconsidered.Theimagescanalsobeassessedusingmoresophisticatedvoxel-basedquantitativemethods,suchasstatisticalparametricmapping(SPM),whichautomaticallymapsbrainregionstoastandardizedatlasandcomparestheimageswithadatabaseofnormalsubjects[77].ComparedwithSPM,computingtheSBPgreatlysimplifiesdataanalysisandoftenprovidesareliablemeasurement.

DynamicSPECTcanalsobeusedtoacquireaseriesofimagesoftraceruptakewithintheROIfromtheinjectiontimetillthetimethetraceriswashedoutorthedistributionreachesequilibrium.Consequentialkineticanalysisoftheacquireddynamicdatausingtissuecompartmentmodelsprovidesuniqueinformationthatimprovesthediscriminationbetweenhealthyanddiseasedtissuecomparedwithstaticimages[25].

FutureperspectivesThemajorchallengefacingthefutureofbrainSPECTimagingisthewidespreadadoptionandavailabilityofPETinclinicalsetting.PEThasbetterspatialresolutionandhighersensitivitythanSPECT.TocompetewithPET,futurebrainSPECTsystemsshouldprovideimageswithbetterqualityandimprovedquantitativeaccuracythatisequivalentoratleastcomparablewithPET.

Software

SoftwaredevelopmentinSPECTiscurrentlyfocusingonimprovingimagequalityandquantitativeaccuracyofreconstructedimages.Thisincludesthedevelopmentofnovelreconstructionalgorithmsandimagecorrectiontechniques.Promisingresultshavebeenachievedusingmaximumaposteriori(MAP)reconstructionalgorithmsthatincorporateanatomicalknowledgeintotheSPECTreconstructionprocess.TheanatomicalinformationcanbeobtainedfromMRimages,whichareroutinelyperformedintheclinicforbrainstudies.MRIprovidesdetailedstructural

Page 16: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 16 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

informationaboutthebrainwhichismissinginSPECT.InMAPreconstruction,theanatomicalinformationcouldbeusedasapriortoreducestatisticalnoiseandimprovetheresolution,especiallywhenanatomicalregion-basedpriorsareusedtoreducenoiseinsideeachstructurebutnotacrosstheboundaries[74,78,79].Differentpriorscanalsobeusedforeachstructurebasedonitsbiologicalproperties.Fig.8.9showsexamplescomparingOS-EMwithMAPreconstructionsusingdifferentregionalpriors.TheOS-EMimagescontainasignificantamountofnoise.IntheMAPreconstruction,auniformpriorwasusedforthestriatumassuminguniformuptake.Asmoothpriorwasusedforthebackgroundtoreducenoise.ImagesfromMAPreconstructionportraymuchreducednoiseandimprovedspatialresolutionforthestriatumcomparedwithOS-EMreconstruction.

Fig.8.9BrainSPECTimagesreconstructedusingdifferentalgorithmsshowingtheactualtracerdistributioninthestriatalbrainphantom(a).OS-EMreconstruction(b).MAPreconstructionwithonlyauniformpriorforthestriatum(c).MAPreconstructionwithasmoothpriorforbackgroundandauniformpriorforthestriatum(d).

Hardware

AmorefundamentalwayofimprovingSPECTimagesistodevelopnewdetectortechnologythatcouldprovideimageswithhighsensitivityandhighspatialresolution.Detectorsusingsemiconductortechnology,wheredirectphotonconversiondetectorssuchascadmiumtelluridearemostlyused,acquiredataatbetterenergyresolutionthanconventionalgammacameras[80,81].Thiscouldpotentiallyprovideimageswithlessscatterandreducecross-talkindual-isotopeimaging.Developmentinmultimodalityimagingisalsobeingpursued.InSPECT/CT,recentdevelopmentsindual-energyCTscannersmakeitpossibletosegmentsofttissuesinCTimageswithouttheuseofcontrastagents.SimultaneousSPECT/MRIisalsopromisingandshouldenablebothMRandSPECTimagesofpatient’sbraintobeacquiredatthesametime,thusprovidingperfectlyregisteredimages.

AcknowledgementsThisworkwassupportedbytheSwissNationalScienceFoundationundergrantSNSF31003A-149957

References1.AbrahamT,FengJ.(2011).Evolutionofbrainimaginginstrumentation.SeminarsinNuclearMedicine,41,202–19.

Page 17: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 17 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

2.KeyesJ,Jr,OrlandeaN,HeetderksW,etal.(1977).TheHumongotron-ascintillation-cameratransaxialtomograph.JournalofNuclearMedicine,18,381–7.

3.JaszczakRJ,MurphyPH,HuardD,etal.(1977).Radionuclideemissioncomputedtomographyoftheheadwith99mTcandascintillationcamera.JournalofNuclearMedicine,18,373–80.

4.AccorsiR.(2008).Brainsingle-photonemissionCTphysicsprinciples.AJNRAmericanJournalofNeuroradiology,29,1247–56.

5.JaszczakRJ.(2006).Theearlyyearsofsinglephotonemissioncomputedtomography(SPECT):ananthologyofselectedreminiscences.PhysicsinMedicineandBiology,51,R99–115.

6.MadsenMT.(2007).RecentadvancesinSPECTimaging.JournalofNuclearMedicine,48,661–73.

7.HuttonBF.(2010).NewSPECTtechnology:potentialandchallenges.EuropeanJournalofNuclearMedicineandMolecularImaging,37,1883–6.

8.CamargoEE.(2001).BrainSPECTinneurologyandpsychiatry.JournalofNuclearMedicine,42,611–23.

9.CatafauAM.(2001).BrainSPECTofdopaminergicneurotransmission:anewtoolwithprovedclinicalimpact.NuclearMedicineCommunications,22,1059–60.

10.DevousMD.(2002).Functionalbrainimaginginthedementias:roleinearlydetection,differentialdiagnosis,andlongitudinalstudies.EuropeanJournalofNuclearMedicine,29,1685–96.

11.SeoY,MariC,andHasegawaBH.Technologicaldevelopmentandadvancesinsingle-photonemissioncomputedtomography/computedtomography.SeminarsinNuclearMedicine,38,177–98.

12.FrancBL,MyersR,PoundsTR,etal.(2012).ClinicalutilityofSPECT-(low-dose)CTversusSPECTaloneinpatientspresentingforbonescintigraphy.ClinicalNuclearMedicine,37,26–34.

13.BeisterM,KolditzD,andKalenderWA.(2012).IterativereconstructionmethodsinX-rayCT.PhysicsinMedicine,28,94–108.

14.BeekmanF,andvanderHaveF.(2007).Thepinhole:gatewaytoultra-high-resolutionthree-dimensionalradionuclideimaging.EuropeanJournalofNuclearMedicineandMolecularImaging,V34,151–61.

15.CherrySR.(2004).Invivomolecularandgenomicimaging:newchallengesforimagingphysics.PhysicsinMedicine,49,R13–48.

16.MeikleSR,KenchP,KassiouM,etal.(2005).SmallanimalSPECTanditsplaceinthematrixofmolecularimagingtechnologies.PhysicsinMedicineandBiology,2005,22.

17.ZaidiH,(ed.)(2006).Quantitativeanalysisinnuclearmedicineimaging.NewYork:Springer2006.

18.ZaidiH,andElFakhriG.(2008).Isabsolutequantificationofdopaminergicneurotransmissionstudieswith123ISPECTreadyforclinicaluse?EuropeanJournalofNuclearMedicineand

Page 18: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 18 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

MolecularImaging,35,1330–3.

19.RoweRK,AarsvoldJN,BarrettHH,etal.(1993).AstationaryhemisphericalSPECTimagerforthree-dimensionalbrainimaging.JournalofNuclearMedicine,34,474–80.

20.GennaS,andSmithAP.(1988).ThedevelopmentofAspect,anannularsingle-crystalbraincameraforhigh-efficiencySPECT.IEEETransactionsinNuclearSciences,35,654–8.

21.ElFakhriG,OuyangJ,ZimmermanRE,etal.(2006).Performanceofanovelcollimatorforhigh-sensitivitybrainSPECT.MedicalPhysics,33,209–15.

22.RogersWL,ClinthorneNH,ShaoL,etal.(1988).SPRINTII:asecondgenerationsinglephotonringtomograph.IEEETransactionsinNuclearSciences,7,291–7.

23.SeibylJP,StobbartHA,MartinD,etal.(2002).EvaluationofhighresolutionNeuroFocusSPECTdeviceforsmallanimalimaging(abstract).JournalofNuclearMedicine,2002,43

24.FarncombeT,CellerA,NollD,etal.(1999).DynamicSPECTimagingusingasinglecamerarotation(dSPECT).IEEETransactionsinNuclearSciences,46,1055–61.

25.GullbergGT,ReutterBW,SitekA,etal.(2010).Dynamicsinglephotonemissioncomputedtomography-basicprinciplesandcardiacapplications.PhysicsinMedicineandBiology,55,R111–91.

26.WongWH,LiH,UribeJ,etal.(2001).Feasibilityofahigh-speedgamma-cameradesignusingthehigh-yield-pileup-event-recoverymethod.JournalofNuclearMedicine,42,624–32.

27.ActonPD,andMozleyPD.(1999).Singlephotonemissiontomographyimaginginparkinsoniandisorders:areview.BehaviouralNeurology,12,11–27.

28.TatschK,AsenbaumS,BartensteinP,etal.(2002).EuropeanassociationofnuclearmedicineprocedureguidelinesforbrainneurotransmissionSPETusing123I-labelleddopaminetransportligands.EuropeanJournalofNuclearMedicineandMolecularImaging,29,BP30–5.

29.DjangDS,JanssenMJ,BohnenN,etal.(2012).SNMpracticeguidelinefordopaminetransporterimagingwith123I-ioflupaneSPECT1.0.JournalofNuclearMedicine,53,154–63.

30.DevousMD,LoweJL,PayneJK.(1992).Dual-isotopebrainSPECTimagingwithTechnetiumandI-123—Validationbyphantomstudies.JournalofNuclearMedicine,33,2030–5.

31.DevousMD,PayneJK,andLoweJL.(1992).Dual-isotopebrainSPECTimagingwithTechnetiumandI-123—ClinicalvalidationusingXe-133SPECT.JournalofNuclearMedicine,33,1919–24.

32.IvanovicM,WeberDA,LoncaricS,etal.(1994).FeasibilityofdualradionuclidebrainimagingwithI-123andTc-99m.MedicalPhysics,21,667–74.

33.KakAC,andSlaneyM.(1988).Principlesofcomputerizedtomographicimaging.IEEEPress.

34.HudsonHM,andLarkinRS.(1994).Acceleratedimagereconstructionusingorderedsubsetsofprojectiondata.IEEETransactionsinNuclearSciences,13,601–9.

Page 19: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 19 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

35.LangeK,andCarsonR.(1984).EMreconstructionalgorithmsforemissionandtransmissiontomography.JournalofComputerAssistedTomography,8,306–16.

36.CherrySR,SorensonJA,andPhelpsME.(2012).Physicsinnuclearmedicine,4thedn.Philadelphia,PA:ElsevierSaunders.

37.ZaidiH,andKoralKF.(2004).Scattermodellingandcompensationinemissiontomography.EuropeanJournalofNuclearMedicineandMolecularImaging,31,761–82.

38.HuttonBF,BuvatI,andBeekmanFJ.(2011).ReviewandcurrentstatusofSPECTscattercorrection.PhysicsinMedicineandBiology,56,R85–112.

39.MetzCE,AtkinsFB,andBeckRN.(1980).Thegeometrictransferfunctioncomponentforscintillationcameracollimatorswithstraightparallelholes.PhysicsinMedicineandBiology,25,1059–70.

40.TanakaM,UeharaS,KojimaA,etal.(2007).MonteCarlosimulationofenergyspectrafor123Iimaging.PhysicsinMedicineandBiology,52,4409–25.

41.BudingerTF,GullbergGT,HuesmanRH.(1979).Emissioncomputedtomography.Imagereconstructionfromprojections:implementationandapplications.32,147–246.

42.ChangL-T.(1978).Amethodforattenuationcorrectioninradionuclidecomputedtomography.IEEETransactionsinNuclearSciences,NS-25,638–43.

43.ZaidiH,andHasegawaBH.(2003).Determinationoftheattenuationmapinemissiontomography.JournalofNuclearMedicine,44,291–315.

44.TsuiBMW,GullbergGT,EdgertonER,etal.(1989).CorrectionofnonuniformattenuationincardiacSPECTimaging.JournalofNuclearMedicine,30,497–507.

45.TsuiBMW,ZhaoXD,FreyEC,etal.(1994).Quantitativesingle-photonemissioncomputed-tomography—basicsandclinicalconsiderations.SeminarsinNuclearMedicine,24,38–65.

46.JaszczakRJ,GreerKL,FloydCE,etal.ImprovedSPECTquantificationusingcompensationforscatteredphotons.JournalofNuclearMedicine,25,893–900.

47.IchiharaT,OgawaK,MotomuraN,etal.(1993).Comptonscattercompensationusingthetriple-energywindowmethodforsingle-anddual-isotopeSPECT.JournalofNuclearMedicine,34,2216–21.

48.KingMA,HademenosGJ,andGlickSJ.(1992).Adual-photopeakwindowmethodforscattercorrection.JournalofNuclearMedicine,33,605–12.

49.BuvatI,BenaliH,ToddpokropekA,etal.(1994).Scattercorrectioninscintigraphy—thestate-of-the-art.EuropeanJournalofNuclearMedicine,21,675–94.

50.ElFakhriG,MaksudP,KijewskiMF,etal.(2000).Scatterandcross-talkcorrectionsinsimultaneousTc-99m/I-123brainSPECTusingconstrainedfactoranalysisandartificialneuralnetworks.IEEETransactionsinNuclearSciences,47,1573–80.

51.FloydCE,Jr,JaszczakRJ,GreerKL,etal.(1985).DeconvolutionofComptonscatterinSPECT.

Page 20: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 20 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

JournalofNuclearMedicine,26,403–8.

52.GillandDR,TsuiBMW,McCartneyWH,etal.(1988).DeterminationoftheoptimumfilterfunctionforSPECTimaging.JournalofNuclearMedicine,29,643–50.

53.YanchJC,FlowerMA,andWebbS.(1988).AcomparisonofdeconvolutionandwindowedsubtractiontechniquesforscattercompensationinSPECT.IEEETransactionsinNuclearSciences,7,13–20.

54.FreyEC,andTsuiBMW.(1996).Anewmethodformodelingthespatially-variant,object-dependentscatterresponsefunctioninSPECT.In:1996IEEENuclearScienceSymposiumConferenceRecord,pp.1082–6.Anaheim,CA.

55.DuY,TsuiBMW,andFreyEC.(2006).Model-basedcompensationforquantitativeI-123brainSPECTimaging.PhysicsinMedicineandBiology,51,1269–82.

56.CaoZJ,FreyEC,andTsuiBMW.(1994).AscattermodelforparallelandconvergingbeamSPECTbasedontheKlein–Nishinaformula.IEEETransactionsinNuclearSciences,41,1594–600.

57.RiaukaTA,andGortelZW.(1994).Photonpropagationanddetectioninsingle-photonemissioncomputedtomograph—ananalyticalapproach.MedicalPhysics,21,1311–22.

58.BeekmanFJ,deJongHWAM,andvanGelovenS.(2002).Efficientfully3-DiterativeSPECTreconstructionwithMonteCarlo-basedscattercompensation.IEEETransactionsinMedicalImaging,21,867–77.

59.LazaroD,ElBitarZ,BretonV,etal.(2005).Fully3DMonteCarloreconstructioninSPECT:afeasibilitystudy.PhysicsinMedicineandBiology,50,3739–54.

60.ElFakhriG,MooreSC,MaksudP,etal.(2001).AbsoluteactivityquantitationinsimultaneousI-123/Tc-99mbrainSPECT.JournalofNuclearMedicine,42,300–8.

61.ErlandssonK,VisvikisD,WaddingtonWA,etal.(1999).DualradionuclidebrainimagingwithTc-99mandI-123withcrosstalkcorrectionsusingmultipleenergywindows.EuropeanJournalofNuclearMedicine,26,1186.

62.KnesaurekK,andMachacJ.(1997).Enhancedcross-talkcorrectiontechniqueforsimultaneousdual-isotopeimaging:aTl-201/Tc-99mmyocardialperfusionSPECTdogstudy.MedicalPhysics,24,1914–23.

63.LinksJM,PrinceJL,andGuptaSN.(1996).AvectorWienerfilterfordual-radionuclideimaging.IEEETransactionsinMedicalImaging,15,700–9.

64.NeumannDR,andChenEQ.(1992).Convolution-basedmethodofenergyspectralcrosstalkcorrectionforsimultaneous,dual-isotopeSPECT.Radiology,185,176.

65.YangJT,YamamotoK,SadatoN,etal.(1997).Clinicalvalueoftriple-energywindowscattercorrectioninsimultaneousdual-isotopesingle-photonemissiontomographywith123I-BMIPPand201Tl.EuropeanJournalofNuclearMedicine,24,1099–106.

66.DuY,TsuiBMW,andFreyEC.(2007).Model-basedcrosstalkcompensationforsimultaneous

Page 21: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 21 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

Tc-99m/I-123dual-isotopebrainSPECTimaging.MedicalPhysics,34,3530–43.

67.PretoriusPH,KingMA,PanTS,etal.(1988).ReducingtheinfluenceofthepartialvolumeeffectonSPECTactivityquantitationwith3Dmodellingofspatialresolutioniniterativereconstruction.PhysicsinMedicineandBiology,43,407–20.

68.SoretM,RiddellC,HapdeyS,etal.(2001).SimultaneousattenuationandpartialvolumeeffectcorrectionforSPECTI-123dopaminereceptorimaging.JournalofNuclearMedicine,42,57p.

69.RoussetO,RahmimA,AlaviA,etal.(2007).PartialvolumecorrectionstrategiesinPET.PETClinics,2,235–49.

70.ErlandssonK,BuvatI,PretoriusPH,etal.(2012).Areviewofpartialvolumecorrectiontechniquesforemissiontomographyandtheirapplicationsinneurology,cardiologyandoncology.PhysicsinMedicineandBiology,57,R119–59.

71.MeltzerCC,ZubietaJK,LinksJM,etal.(1994).MRI-basedcorrectionforPETpartialvolumeeffectsinthepresenceofheterogeneityingray-matterradioactivity.JournalofNuclearMedicine,35,P197.

72.RoussetOG,MaY,EvansAC.(1998).CorrectionforpartialvolumeeffectsinPET:principleandvalidation.JournalofNuclearMedicine,39,904–11.

73.DuY,TsuiBMW,FreyEC.(2005).PartialvolumeeffectcompensationforquantitativebrainSPECTimaging.IEEETransactionsinMedicalImaging,24,969–76.

74.BaeteK,NuytsJ,VanLaereK,etal.(2004).EvaluationofanatomybasedreconstructionforpartialvolumecorrectioninbrainFDG-PET.NeuroImage,23,305–17.

75.IidaH,LawI,PakkenbergB,etal.(2000).QuantitationofregionalcerebralbloodflowcorrectedforpartialvolumeeffectusingO-15waterandPET:I.Theory,erroranalysis,andstereologiccomparison.JournalofCerebralBloodFlowMetabolism,20,1237–51.

76.ActonPD,MeyerPT,MozleyPD,etal.(2000).Simplifiedquantificationofdopaminetransportersinhumansusing[99mTc]TRODAT-1andsingle-photonemissiontomography.EuropeanJournalofNuclearMedicine,27:1714–18.

77.HabrakenJB,BooijJ,SlomkaP,etal.(1999).Quantificationandvisualizationofdefectsofthefunctionaldopaminergicsystemusinganautomaticalgorithm.JournalofNuclearMedicine,40,1091–7.

78.BaeteK,NuytsJ,VanPaesschenW,etal.(2004).Anatomical-basedFDG-PETreconstructionforthedetectionofhypo-metabolicregionsinepilepsy.IEEETransactionsinMedicalImaging,23,510–19.

79.NuytsJ,BaeteK,BequeD,etal.(2005).ComparisonbetweenMAPandpostprocessedMLforimagereconstructioninemissiontomographywhenanatomicalknowledgeisavailable.IEEETransactionsinMedicalImaging,24,667–75.

80.OgawaK,OhmuraN,IidaH,etal.(2009).Developmentofanultra-highresolutionSPECTsystemwithaCdTesemiconductordetector.AnnalsofNuclearMedicine,23,763–70.

Page 22: Oxford Medicine Online Principles of brain single-photon emission computed tomography ... · 2018. 10. 3. · Principles of brain single-photon emission computed tomography imaging

Principles of brain single-photon emission computed tomography imaging

Page 22 of 22

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). ©Oxford University Press, 2015. All RightsReserved. Under the terms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in OxfordMedicine Online for personal use (for details see Privacy Policy). Subscriber: Habib Zaidi; date: 27 November 2015

81.TakahashiY,MiyagawaM,NishiyamaY,etal.(2013).PerformanceofasemiconductorSPECTsystem:comparisonwithaconventionalAnger-typeSPECTinstrument.AnnalsofNuclearMedicine,27,11–16.