DKI (弥散峰度成像)ppt课件

DiffusionalKurtosisImaging,DKI,1,Contents,DWI(diffusionweightedimaging),DTI(diffusiontensorimaging),DKI(diffusionkurtosisimaging),2,DWI原理,MR图像的信号组织T1、T2驰豫时间、H1的密度、分子弥散运动DWI图像利用扩散敏感梯度脉冲将水分子弥散效应扩大，来研究不同组织中水分子扩散运动的差异,3,DWI评估弥散的参数,通过两个以上不同弥散敏感梯度值（b值）的弥散加权象，可计算出弥散敏感梯度方向上水分子的表观弥散系数（apparentdiffusioncoefficientADC）ADC=In(S低/S高)/(b高-b低)弥散敏感系数(b)值=r22g2(-/3)b值的取值范围为010000s/mm2，较大的b值具有较大的弥散权重，对水分子的弥散运动越敏感，并引起较大的信号下降，但b值越大，图像信噪比也相应下降，如果b值太小，易受T2加权的影像，产生所谓的T2透射效应(T2shinethrougheffect)，一般来说用大b值差的图像测得的ADC值较准确，故侧ADC值时宜选较高b值和较大的b值差ADC反映了水分子的扩散运动的能力,指水分子单位时间内扩散运动的范围，越高代表水分子扩散能力越强。,4,均质介质中可以水分子的自由运动为各向同性，即在各个方向上的弥散强度大小一致，弥散张量D描述为球形，沿磁共振的三个主坐标的特征值为1=2=3在脑白质中由于髓鞘的阻挡，水分子的弥散被限制在与纤维走行一致的方向上，具有较高的各向异性，此时弥散张量可表示为椭球形，其特征值123，最大特征值对应的方向与经过该体素的纤维束走行平行,5,defectsofDTI,ConventionalDTIfailstofullyutilizetheMRdiffusionmeasurementsthatareinherenttotissuemicrostructure.DTIcomputesapparentdiffusivitybasedontheassumptionthatdiffusionweighted(DW)MRsignalhasamonoexponentialdependenceonthediffusionfactor(b-value).DTIimplicitlyassumesthatwatermoleculediffusionoccursinafreeandunrestrictedenvironmentwithaGaussiandistributionofdiffusiondisplacement.,6,defectsofDTI,Inbiologicaltissue,complexcellularmicrostructuresmakewaterdiffusionahighlyhinderedorrestrictedprocess.Non-monoexponentialdecaysareexperimentallyobservedinbothwhitematterandgraymatter.Moreover,thesimplifieddescriptionofthediffusionprocessinvivobya2nd-order3DdiffusivitytensorpreventsDTIfrombeingtrulyeffectiveincharacterizingrelativelyisotropictissuesuchasGM.EveninWM,theDTImodelcanfailifthetissuecontainssubstantialcrossingordivergingfibers.,7,defectsofDTI,Asaresult,DTIquantitationisb-valuedependentandDTIfailstofullyutilizethediffusionmeasurementsthatareinherenttotissuemicrostructure.,8,Kurtosis,Kurtosisherereferstotheexcesskurtosisthatisthenormalizedandstandardizedfourthcentralmomentofthewaterdisplacementdistribution.ItisadimensionlessmeasurethatquantifiesthedeviationofthewaterdiffusiondisplacementprofilefromtheGaussiandistributionofunrestricteddiffusion,providingameasureofthedegreeofdiffusionhindranceorrestriction.fourthcentralmoment:四阶中心距，主要用来衡量随机分布变量的分布在均值附近的陡峭程度SincethedeviationfromGaussianbehaviorisgovernedbythecomplexityofthetissuewithinwhichthewaterisdiffusing,thisexcessdiffusionalkurtosiscanberegardedasameasureofatissuesdegreeofstructure.,9,OtheradvantagesofDKI,Meankurtosis(MK),theaverageapparentkurtosisalongalldiffusiongradientencodingdirections,hasbeenmeasuredanddemonstratedtoofferanimprovedsensitivityindetectingdevelopmentalandpathologicalchangesinneuraltissuesascomparedtoconventionalDTI.Inaddition,directionalkurtosisanalysishasbeenformulatedtorevealdirectionallyspecificinformation,suchasthewaterdiffusionkurtosesalongthedirectionparallelorperpendiculartotheprinciplewaterdiffusiondirectionasdeterminedbythe2nd-orderdiffusiontensor,10,DKIprovidesahigher-orderdescriptionofrestrictedwaterdiffusionprocessbya2nd-order3Ddiffusivitytensor(DTasinconventionalDTI)togetherwitha4th-order3Dkurtosistensor(KT).,11,Conditions,Themethodisbasedonthesametypeofpulsesequencesemployedforconventionaldiffusion-weightedimaging(DWI),buttherequiredbvaluesaresomewhatlargerthanthoseusuallyusedtomeasurediffusioncoefficients.Inthebrain,bvaluesofabout2000s/mm2aresufficient.Atleast15non-collinearandnon-coplanardirectionsarerequiredtoconstructKT.,12,DKIvsq-spaceimagingtechniques,DKIhasacloserelationshiptoq-spaceimagingtechniques.q-spaceimagingmethodshaveindeedrecentlybeenemployedtoestimatediffusionalkurtosis.Theprincipaldifferencebetweenthemisthatq-spaceimagingseekstoestimatethefulldiffusiondisplacementprobabilitydistributionratherthanjustthekurtosis.Asaconsequence,q-spaceimagingismoredemandingintermsofimagingtimeandgradientstrengths.MeasuringthediffusionalkurtosisrequiresonlymodestincreasesinbvaluesAndDKIislessdemandingintermsofhardwarerequirementsandpostprocessingeffort.,13,Kurtosistensor(KT)derivedparameters,MK(meankurtosis):MKisameasureoftheoverallkurtosis.Itdoesnothaveanydirectionalspecificity.MK的大小取决于感兴趣区内组织的结构复杂程度，结构越复杂非正态分布水分子扩散受限越显著，MK也即越大K(Axialkurtosis)andK(Radialkurtosis):canbedefinedasthekurtosisparallelandperpendiculartotheprinciplediffusioneigenvector(e1)K越大表明在该方向非正态分布水分子扩散受限越明显，反之则表明扩散受限越弱FAK(fractionalanisotropyofkurtosis)SimilartoFAinDTI,theanisotropyofdirectionalkurtosiscanbeconvenientlydefinedasFAKKA越小即表示越趋于各向同性扩散;若组织结构越紧密越规则，KA越大,14,DKIparametricmaps,15,DKIparametricmaps,TypicalDKI-derivedparametricmapsfromasinglesliceofa)invivo,b)formalin-fixedadultratbrainsandc)anormalhumansubject(male,44yearsold).Axialdiffusivity(/),radialdiffusivity(),meandiffusivity(MD),axialkurtosis(K/),radialkurtosis(K),meankurtosis(MK),fractionalanisotropy(FA),directionallyencodedcolourFA(DEC-FA)andfractionalanisotropyofkurtosis(FAK)mapsarecomputedfromDKImodel.,16,DKIparametricmaps,For(a),rawDWIswereacquiredbySEEPIwithTR/TE=3000/30.3ms,/=5/17ms,slicethickness=1mm,FOV=3030mm2,datamatrix=128128(zerofilledto256256),NEX=4,6b-values(0.0,0.5,1.0,1.5,2.0and2.5ms/m2)andalong30directionsusing7Tscanner,17,DKIparametricmaps,For(b),rawDWIswereacquiredwiththesameparametersasthoseforinvivoexceptTE=34.3ms,=9msandb-valuesof0.0,1.0,2.0,3.0,4.0and5.0ms/m2.Alargerb-valuerangewasusedinexvivoexperimentduetothegenerallylowerdiffusivities.,18,DKIparametricmaps,For(c),rawDWIswereacquiredbySEEPIwithTR/TE=2300/109ms,slicethickness=2mm,FOV=256256mm2,datamatrix=128128,NEX=2,6b-values(0.0,0.5,1.0,1.5,2.0and2.5ms/m2)andalong30directionsusinga3TSiemensscanner,19,DKIparametricmaps,HigherMKisfoundinWM,indicatingagenerallyhigherdegreeofdiffusioncomplexityandrestrictionintheWMstructures.ItcanbeseenfromthedirectionalkurtosismapsthatsuchhighMKinWMismainlycontributedbyK.Thissuggeststheexistenceofheterogeneityandrestricteddiffusioninaxonalstructures,BothMKandKexhibitstrongcontrastbetweenWMandGMstructures.,20,DKIparametricmaps,BothMKandKexhibitstrongcontrastbetweenWMandGMstructures.PositivemeananddirectionalkurtosesareobservedinbothWMandGM,indicatingfasterDWsignaldecayatlowerb-valuesandrestricteddiffusionenvironmentinbothWMandGMunderinvivoandformalin-fixedconditions.,21,DKIshowsageneraldecreaseindiffusivityandincreaseinkurtosisinWMandGMofthefixedbrains,Thebreakdownsofneurofilamentsandmicrotubulescausedbyfixativesarebelievedtoproducemorediffusionbarriersandhenceleadtothe/decreaseandK/increase.Otherfixationeffectssuchastissueshrinkage,decreaseinmembranepermeability,increaseinaxonalpackingdensityandreductionofextracellularspaceinparenchymaalsolikelycontributetothesignificantdecreaseandKincrease.,22,Directionalkurtosisanalysisoffixedexperimentalautoimmuneencephalitis(EAE)spinalcord,TheinflammatoryneurodegenerativediseaseEAEischaracterizedbybothaxonallossanddemyelinationInrecentDKIstudies,therearepromisingresultsofusingMKtodetectchangesinnormalorpathologicalneuraltissueHowever,asanaverageofkurtosesalongallthediffusiondirections,MKcanlosesensitivityandspecificityinprobingdirectionalchangesofpathologicaltissue,23,EAEspinalcord,K/isfoundtobesignificantlyincreasedand/decreasedinthelesionarea/reductionislikelyduetocytoskeletalperturbationordebrisformationwhentheaxonalstructuresbreakdownInaddition,increaseswhereasKdecreaseslikelybecauseofthedemyelinationandaxonallossthatalsoleadtolessdiffusionrestrictioninradialdirection.,24,EAEspinalcord,ThedirectionallyaveragedMDandMKarefoundtobelesssensitivetoEAEpathologyduetotheoppositetrendsofdiffusivityandkurtosischangesinaxialandradialdirection.,25,MonitoringpostnatalbrainmaturationbyconventionalDTI,CC:corpuscallosum（胼胝体）;EC:externalcapsule（外囊）;CP:cerebralpeduncle（大脑脚）;AC:anteriorcommissure;（前联合）CT:cerebralcortex（脑皮质）;HP:hippocampus（海马）;CPu:caudateputamen（新纹状体）,26,MonitoringpostnatalbrainmaturationbyconventionalDTI,Thesensitivityof/indetectingratbrainWMmaturationisgenerallyobservedtobethehighestatlowb-valueAtrelativelylowb-values,theapparentdiffusivityisprimarilycontributedfromthefastwaterdiffusionactivitiesinextracellularspacethatdependonbothcellularmicrostructureandmembranepermeability.Theuseoflowb-valuecanbestdetectthesechanges.Thehigh/sensitivityatlowb-valueobservedinthecurrentstudysuggeststhealterationsofthesefastwaterdiffusionactivitiesalongaxonaldirectionduringbrainmaturation.Suchalterationsmayresultfromtheincreaseinpackingdensityoffiberbundlesandaxons,axonaldiameterincrease,changesinneurofibrils,andincreasedcomplexityofextracellularmatrix.,27,MonitoringpostnatalbrainmaturationbyconventionalDTI,Whereasthatofisthehighestathighb-valueThediffusionchangesprobedinWMusinghighb-valuesareascribedmoretotheslowwatermoleculediffusionparticularlyalongtheradialdirectionwhentraversingthemembranesandmyelinsheathsThehighsensitivityofathighb-valueindetectingbrainmaturationshowninthefigurelikelyreflectstheseWMmicrostructuralchanges,includingmyelinationandaxonaldensityanddiameterchangesduringpostnatalbraindevelopment.,28,MonitoringpostnatalbrainmaturationbyconventionalDTI,FAquantitationisalsoaffectedbytheb-valueanditsabilityindetectingbrainmaturationalchangesvariesamongdifferentstructures.,29,MonitoringpostnatalbrainmaturationbyDKI,Figure7ashowsthatthesensitivityoffittingallthemulti-b-valueDWIstoDTImodelisgenerallysimilartothatofemployingamediumb-value(b=1.5ms/m2)showninFigure6InFigure7b,thegeneralandcontinualkurtosisincreasewithageisobserved,indicatingthatmorediffusionrestrictionoccursduringbrainmaturationinbothWMandGMstructures.TheDKI-deriveddiffusivityandkurtosisindicesarehighlysensitivetobraindevelopmentalchanges.,30,MonitoringpostnatalbrainmaturationbyDKI,Both/andK/ofWMarefoundtoincreaseswithage,whichmayarisefromvariousbiologicaleventsduringearlypostnatalbrainmaturation.Theincreaseofdiffusivitycanbecausedbyaxoplasmicflowduringthemyelinationperiodneuronallossandaxonalpruningthatshortenstheaxonlengthcanleadtoanincreaseofrestriction,31,MonitoringpostnatalbrainmaturationbyDKI,TheincreaseofKinWMislikelyascribedtothemyelinationandmodificationofaxonalstructuresthatincreasesrestrictionintheradialdirection.DKIanalysisalsorevealsthatdiffusionrestrictionintherelativelyisotropicGMincreaseswithage.ThismayreflectthemoredenselypackedstructuresandthedendriticarchitecturalmodificationinGM,32,DTIVSDKIinmonitoringpostnatalbrainmaturation,WhenthereisalargeK,theestimateddiffusivityinconventionalDTIshowsalargediscrepancywiththediffusivityestimatedinDKI