会员注册 | 登录 | 微信快捷登录 QQ登录 微博登录 | 帮助中心 人人文库renrendoc.com美如初恋!
站内搜索 百度文库

热门搜索: 直缝焊接机 矿井提升机 循环球式转向器图纸 机器人手爪发展史 管道机器人dwg 动平衡试验台设计

   首页 人人文库网 > 资源分类 > PDF文档下载

外文资料-- Glioma Tissue Modeling by Combing the Information of MRI and in vivo Multivoxel MRS.PDF

  • 资源星级:
  • 资源大小:1.62MB   全文页数:4页
  • 资源格式: PDF        下载权限:注册会员/VIP会员
您还没有登陆,请先登录。登陆后即可下载此文档。
  合作网站登录: 微信快捷登录 支付宝快捷登录   QQ登录   微博登录
友情提示
2:本站资源不支持迅雷下载,请使用浏览器直接下载(不支持QQ浏览器)
3:本站资源下载后的文档和图纸-无水印,预览文档经过压缩,下载后原文更清晰   

外文资料-- Glioma Tissue Modeling by Combing the Information of MRI and in vivo Multivoxel MRS.PDF

GliomaTissueModelingbyCombingtheInformationofMRIandinvivoMultivoxelMRSWeibeiDOU,AoyanDONG,PingCHITsinghuaNationalLaboratoryforInformationScienceandTechnologyDept.ofElectronicEngineering,TsinghuaUniversity,Beijing,100084,P.R.Chinaemaildouwbtsinghua.edu.cnShaowuLINeuroimagingCenterofTiantanHospitalCapitalMedicalUniversity,Beijing,P.R.ChinaJeanMarcCONSTANSUnitédIRM,EA3916,CHRUCaen,FranceAbstractThispaperpresentsagliomamodelizationmethodandaregressionlikemodeltocreateagraduallygliomaimageGlioIm.Multimodalsignal,imagesofmagneticresonanceimagingMRIandinvivomultivoxelMRspectroscopyMRSarecombinedbytheregressionlikemodelwithspatialresolutionregistration.ThismodelingmethodconsistsoffeaturemodelsofgliomasuchasthesignalintensityofMRimageandthemetabolitechangesofMRS,thecorrelationmodelnotedasmetabolitesratioMetaRandthecombinedregressionlikemodel.TheestimatedGlioImincludesbothbrainstructureandgliomagradeinformation.Anonlinearmodelisproposedandvalidatedinthispaper.ThetestingdataisacquiredbySiemensTrioTim3TandSyngoMRB15atBeijingTiantanhospitalofChina.TheMRSofthreegliomapatients,twoaffectedbyastrocytomaandonebyglioma,andthechemicalshiftimagingCSIreferenceT2imageswereconsideredinourvalidationexperiment.TheresultingGlioImsarecomparedwithgroundtruthprovidedbyneuroradiologistsofTiantanandverifiedwiththeirpathologyreport.Theyreportthatourmethodandmodelareveryefficient.KeywordsMRSpectroscopybraingliomachemicalshiftimagingMRIimagemodelingcombinationI.INTRODUCTIONTodiagnosebraintissueabnormalities,liketumor,itsnecessarytousemultispectralmagneticresonanceimagesMRIs,suchasT1weight,T2weight,Gadolinium,FLAIRetcinordertofindsomeoftumorspropertiessuchassize,position,sort,andrelationshipwithothertissues,etc...Butthetumortypeandgradeareusuallydiagnosedfromhistopathologicalexaminationofasurgicalspecimen.However,Hydrogen11HmagneticresonancespectroscopyMRSisanoninvasiveMRtechniquethatprovidesbiochemicalinformationofmetabolites.Themajorbiochemicalcharacteristicscannoninvasivelyprovideusefulinformationonbraintumortypeandgrade1.Inmanystudies,invivo1HMRShasbeenpresentedfordeterminingthetypeandgradeoftumors123.SinceinvivoMRSmeasurementsandanalysisaredependentontheacquisitiontechnicalthatcompromisethespatialresolutionandaccuracyforresultingmetabolitevalues4,metabolicchangeswithdiseaseisfrequentlysubtleanddiffuse.Furthermore,bychemicalshiftimagingCSItechnique,themetaboliteimagessocalledMRspectroscopicimagingMRSIcanbecreatedbymultivoxelMRSinformation,butitisnotvisuallyinterpretableinthesenseofastructuralMRI4.Sothat,forthetumortissueclassification,itisimportantthatMRSIiscombinedwithMRItoestimatethevariationofmetabolitesandtoyieldmuchinformationregardingtissue.Duringmorethanadecade,automaticbraintumorclassificationbyMRShasbeendeveloped5,butthemorecleardefinitionofbraintumortypeandgrademaybeobtainedbycombinationofMRSIandMRI5.AtechniquetodifferentiateglioblastomafrommetastasislesionsbyusingMRIandMRSdatahasbeenpublishedin6.Wangetal.describedaclassificationofbraintumorsbyusingfeaturesselectionandfuzzyconnectednessin7,thesefeaturesareextractedfromMRIandMRSdata.TherearetwodifficultiesforcombingMRSIdataandMRIdatafirstly,thesedataarefromdifferentmodalities,sotheyarenotinthesamespatialresolution,verylowspatialresolutioninvoxelforMRSIandhighspatialresolutioninpixelforMRI.Secondly,oneMRimagecorrespondstothedistributionofalltissues,ortissuestructure.ButoneMRSimageisaprojectionimagewhichcorrespondstoonemetaboliteorratiobetweenseveralmetabolites.SothedifferentmetabolitevaluesmakevariationMRSimages,justlikethemappingofmetabolitedistributionsbyMRSIpresentedin8.ThequestionforapplicationishowtocombinetheseMRSimagesandMRimagestogiveanautomatictissueclassificationresult.ThekeypointofthecombinationishowtomodelthemetabolitedistributionfromMRS,whichcorrespondstoinformationfromMRimages.Forautomaticdescriptionofbraintumortypeandgrade,weproposeamodelizationmethodofgliomatissuesbycombingtheinformation,fromMRimagesandMulitivoxelMRSdata.ItcancreateaMRSweightedMRimageautomaticallywhichkeepsthehighspatialresolutionlikeMRimageandthegreylevelscorrespondtothedeteriorationofbraintissues.ThesecondpartofthispaperintroducesthegliomatissuefeaturesbothinMRSvaluesandinMRimages.Thecombinationmodelingofthetwotypesofinformationispresentedinthethirdsectionanditsvalidationisshowninthefourthsection.Theconclusionaboutourresearchisgivenattheendofthispaper.ThisworkisfundedbyTsinghuaNationalLaboratoryforInformationScienceandTechnology(TNList)CrossdisciplineFoundation9781424447138/10/25.00©2010IEEEII.FEATURESMODELOFGLIOMATISSUEFollowingtheresearchofdiagnosingbraintumorbyMRimagesandMRS,wecansummarizetwotypesofcharacteristicsofglioma,oneisthesignalintensityofT1weightandT2weightimages,andtheotheroneisthechemicalshiftvaluesofmetabolitespresentedbyMRSdata.A.SignalIntensityCharacteristicsofMRimagesWehaveproposedsomefuzzymodelingmethodsofdifferenttumorouscerebraltissuesonMRimagesbasedonfusionoftissuefeaturesin91011.TableIdescribesthecharacteristicsofbraintissuesbycreatingagradualityofsignalintensityasafunctionofdifferenttissuesandsequencesofMRI10,whereCSFistheabbreviationofcerebralspinalfluid,GMtheabbreviationofgraymatter,andWMwhitematter.IntableI,theSeqsisshortforSequencesofMRI.Thesymbolpresentsahypersignalitmeansthatthesignalintensityisveryhighandtheimageisverybright.Thesymbolpresentsahyposignal,theintensityisverylowandtheimageisverydark.Thesymbolmeansthatthesignalintensityishigherthanhyposignal,andmeansthatitisdarkerthanhypersignal.meansthatthesignalintensityislowerthanthehyposignal,andmeansthatitisbrighterthanthehypersignal.AnexampleofT1weightedimagenotedasT1,andT2weightedimagenotedasT2areshowninFig.1TABLEI.SIGNALINTENSITYCHARACTERISTICSOFBRAINTISSUESONMRIMAGESSequencesGradualityofsignalintensityCSFGMWMGliomaEdemaNecrosisT1T2abFigure1.OriginalMRIimagesaT1image,bT2imageB.MetaboliteChangesFeaturesofMRSTABLEII.SCALARDESCRIPTIONOFMETABOLITEVALUESMetabolitelevelabsentverylowlittlelowlowmediumlittlehighhighveryhighabbreviationAVLLLLMLHHVHThereareonlyseveralmetaboliteswhichcorrespondtogliomaamongalargenumberofmetabolitesofhumanbody.NacetylasparateNAA,creatineCr,cholineCho,myoinositolmI,lactateLacandfreelipidsLip.ThevariationofthesemetabolitescanbeorderedinascalarformasshowninTableII,wherethescalarorderisabsent,verylow,littlelow,low,medium,littlehigh,high,veryhigh,whichcorrespondtometabolitevaluesfrom0tomaximum.ThemetabolicchangeswithbraintissuesareshowninTableIII.Itisconcludedfrom121314.TABLEIII.METABOLITECHANGESFEATURESOFBRAINTISSUESONMRSMetabolitevariationofmetabolitescorrespondingwithbraintissuesCSFGMWMGliomaEdemaNecrosisNAAVLVHHL/VLMAChoAMLHH/VHLHACrLHHM/LLLAmILMLHHLH/MALipAVLLHLVHLacLHVLAH/LHLHHIII.MODELIZATIONBYCOMBININGMRSWITHMRITheaimofthismodelizationstudyistocreateagraduallygliomaimage,notedasGlioIm,whichincludesbrainstructureandgliomagradeinformation.IfthegliomagradeinformationisconsideredasacorrelationfunctionbetweenMRsignalandpathologicalchanges.WeproposearegressionlikemodeltoestimatetheGlioImfromMRimagesnotedasMRImandmetabolitechanges.A.CorrelationmodelOneofthecorrelationfunctionsismetabolitechangescorrespondingtoglioma.BycombiningtheinformationinTableIandTableIII,wecanrebuildaconclusionTableIVaboutgliomacharacteristicswithrelativequantizationofmetabolitesofTableIII.Therelativequantizationisratiosbetweenmetabolitevalues,suchastheratioofChoandNAAnotedasCho/NAAinTableIV,itiscalledmetabolitesratioMetaR,andTableIViscalledcorrelationmodelinthispaper.TABLEIV.METABOLITESRATIOCHARACTERISTICOFBRAINTISSUESMetabolitevariationofmetabolitescorrespondingwithbraintissuesCSFGMWMGliomaEdemaNecrosisCho/NAAAVLLVHHACho/CrALLHHAmI/CrMLMHHALip/CrAVLVLHMVHLac/CrLHVLAHHHTheMetaRcharacteristicsofglioma,edemaandnecrosisareenhancedandthenormaltissuesarereduced.TheyassortwithsignalintensitycharacteristicsofT2weightedimagedescribedinTableI.B.RegressionlikemodelwithspatialresolutionregistrationNormaly,MetaRisafunctionofvoxeldecidedbyCSIsliceshowninFig.2.Sothat,itisatwodimensionalfunctionnotedasMetaRi,v,whereiisindexofmetaboliteandvistheindexofvoxelcorrespondedwithCSIslice.Asthesamereason,GlioImcanbecreatedasathreedimensionalfunction,notedasGlioImv,p,g,wherepisindexofpixelcorrespondedwithMRIm,andgisthegreylevelofselectedMRimageandcorrespondstop.Infact,MRImisatwodimensionalfunctionnotedasMRImp,g,whereandg∈G,{}1,2,,,,,...TTPDFLAIRGadoDiffusionPerfusionGConsidertwovariables,MRImandGlioIm,MRImisacertainimagelikeT2,GlioImisanestimatedimage.ThecorrelationmodelMetaRcanbeconsideredasonerelationshipbetweenthem.SotheregressionlikemodelforestimatingGlioImfromMRImcanbecreatedasequation1.Im,,,Im,GliovpgMetaRivMRpgΘ1WhereΘnotesanecessaryoperator,andpcorrespondstov.Ifalinearregressiveisnecessary,equation1canberewrittenas2Im,,,Im,,GliovpgMetaRivMRpgMetaRjv2whereiandjindicatedifferentmetabolites.C.NonlinearRegressionlikemodelToavoidmosaiceffects,weproposeanonlinearregressionlikemodelwithspatialresolutionregistrationin3.Im,Im,,exp,.MRpgGliovpgMetaRivMetaRjvT⎡⎤⎢⎥⎣⎦3whereTisatimeconstantcorrespondingtoMRImp,g.AccordingtothecorrelationmodelofTableIV,theLip/CrandLac/Crarespecificfeatureswhicharedependentonthetumorgrade.Sothat,inthemodelofequation2,wehave{},/,/,//,/ijMetaRChoNaaChoCrmICrLipCrLacCr∈∈IJIJIJ∪,BecausetheJofMetaRisthegrademarker,ittakesaninterceptiveroletomakeadifferentgreylevelfromothervoxelsandindicatesavariablegrade.IV.VALIDATIONANDRESULTA.MaterielThreegliomapatients,twoaffectedbyastrocytomaandonebyglioma,wereconsideredinourvalidationexperiment.ThetestingdataareadatapairconsistedofCSIrawdataandtheirreferenceimages.ThesedatawereacquiredwithSTEAMsequenceatBeijingTiantanhospitalChina,bySiemensMRTrioTim3TandsyngoMRB15.TheMRSrawdataaremeasuredbycsi_st/90protocolwithTR3000/TE72/TM6.T2weightedimagesaremeasuredbyt2_tse_traprotocolwithTR4500/TE80.TwoexamplesofthesedataareshowninFig.2.Thenonlinearregressionlikemodel3isvalidatedbyourtestingexperimentation.MRImof3isT2with0.570.57mm2pixelsizeand5mmslicethickness.ThetimeconstantTin3isindicatedbyhistogrampeakofCSIreferenceimagesinT2.ThemetabolitevaluesarecalculatedbyTHUMRSv0.5developedbyourresearchgroupandpublishedin15.TheCSIslicesnotethattheMRSvoxelsizeis141420mm3.abFigure2.ExampleofCSIslicedownleftwithitsreferenceimagesandmetabolitesvaluescorrespondedwithvoxelsize141420mm3.afromanastrocytomapatient,masculine30yearsold.bfromagliomapatient,feminie48yearsold.B.ResultThevalidationresultscorrespondedtoVOIareshowninFig.3fandFig.4f.Thehighersignalorbrighterpixelinfmarksgreaterpossibilityofgliomaorhighertumorgrade.InFig.3and4,aaretheoriginalT2weightedimageswiththesignofVOI,barethehandlabelresultsasGroundtruthfromneuroradiologistsofTiantan,careonepartofainVOI,daretheresultsofexponentialcomponentofequation3whichpresentsthecombinedinformationofT2andCho/Naa,earetheresultsofsuperpositionofT2andLipLc/Cr.aT2VOIbGroundtruthcOriginalT2inVOIdMetaRCho/NaaeMetaRLacLip/CrfResultingGlioImFigure3.ResultingGlioImfofthepatientaffectedbyastrocytomaC.DiscussionThebrighterpixelinFig.3dorfdenotesnotonlyhigherCho/NaabutalsobrighterT2.BecauseMetaRvaluesinTableIVareconsistentwiththeintensityofT2.Soitmayindicategliomaandhighergraderegion.ThedarkerpixelspresentlowerCho/NaaanddarkerT2,mayindicatenormaltissues.Thentherearesomebrighterpixelsindandf,theyarenotconnectedwithgliomaregion,theyareCSFperhaps,becauseCSFisbrighterinT2.WecanremovethemsimplybyusingregisteredFLAIRimage.TheregisteredGadoliniumimagealsocanbeusedtoindicateenhancedpixelsorregion.aT2VOIbGroundtruthcoriginalT2inVOIdMetaRCho/NaaeMetaRLacLip/CrfResultingGlioImFigure4.ResultingGlioImfofthepatientaffectedbygliomaBecauseamongthe5metaboliteratiosinTableIV,onlythreepresentevidentchanges,likeCho/Naa,Lip/CrandLac/Cr.Theothertworatiosarenotutilizedinourexperiment.ItispossibletouseotherMRIsequencessuchasT1,butitisnecessaryeithertotransformgreylevelsofimageortoinversethevalueofMetaR.Asmentionedin16,animageresultedfromfusionofgliomafeaturesextractedfrommultimodalitysignal,aspresentedin9,canalsobeusedasMRIminthisgliomamodel.V.CONCLUSIONAdvantagesofMRItechniqueprovidemorepossibilitywithmultisequencesandmultimodalitiessignaltorealizethetumordiagnosis,treatmentandprognosis.Butitisheavyworkforprocessingallsignalstodoafinaldecision.SoAutomaticquantificationandcombinationanalysisisveryimportantandthemodelingoftumorfeaturesisthekeypointforperformingit.Wehaveproposedaframeworkoffuzzyfeaturesfusionsystemin16andpublishedsomeresearchresultsaboutfusingthetumorfeaturesextractedfromT1,T2andprotondensityimages9.Inthispaper,wepresenttheprimarystudyaboutthetumorfeaturescombinationofMRSandMRimages.Theproposedmodelingmethodandnonlinearregressionlikemodelarevalidforseparatingthebraintissuesespeciallyglioma.Itwillbeusedfortumortissuesclassification,segmentation,tumortypeandgradedecision,etc.Thereisstillmuchworktoimprovethismodelandtointegrateitwiththefusionsysteminthefuture.REFERENCES1HoweFA,BartonSJ,CudlipSA,StubbsM,SaundersDE,MurphyM,WilkinsP,OpstadKS,DoyleVL,McLeanMA,BellBA,GriffithsJR.Metabolicprofilesofhumanbraintumorsusingquantitativeinvivo1Hmagneticresonancespectroscopy.MagnResonMed.2003Feb49222332.2PreulMC,CaramanosZ,CollinsDL,VillemureJG,LeblancR,OlivierA,PokrupaR,ArnoldD.Accurate,noninvasivediagnosisofhumanbraintumorsbyusingprotonmagneticresonancespectroscopy.NatMed19962323–325.3MajósC,AguileraC,CosM,CaminsA,CandiotaAP,DelgadoGoñiT,SamitierA,CastañerS,SánchezJJ,MatoD,AcebesJJ,ArúsC.InvivoprotonmagneticresonancespectroscopyofintraventriculartumoursofthebrainEurRadiol.2009Aug198204959.4A.A.Maudsley,C.Domenig,V.Govind,A.Darkazanli,C.Studholme,K.Arheart,C.Bloomer,MappingofbrainmetabolitedistributionsbyvolumetricprotonMRspectroscopicimagingMRSIMagneticResonnanceinMedicin615485592009.5GarciaGomezJ.,LutsJ.,JuliaSapeM.,KrooshofP.,TortajadaS.,VicenteJ.,MelssenW.,FusterGarciaE.,OlierI.,PostmaG.,MonleonD.,MorenoTorresA.,PujolJ.,CandiotaA.P.,MartinezBisbalM.C.,SuykensJ.A.K.,BuydensL.,CeldaB.,VanHuffelS.,ArusC.,RoblesM.,Multiprojectmulticenterevaluationofautomaticbraintumorclassificationbymagneticresonancespectroscopy,MagneticResonanceMaterialsinPhysics,BiologyandMedicine,vol.22,Feb.2009,pp.518.6LutsJ.,LaudadioT.,MartinezBisbalM.C.,VanCauterS.,MollaE.,PiquerJ.,SuykensJ.A.K.,HimmelreichU.,CeldaB.,VanHuffelS.,DifferentiationbetweenbrainmetastasesandglioblastomamultiformebasedonMRI,MRSandMRSI,inProc.ofthe22ndIEEEInternationalSymposiumonComputerBasedMedicalSystemsCBMS,Albuquerque,NewMexico,Aug.2009,pp.18.7QiangWang,EiriniKaramaniLiacouras,EricksonMiranda,UdayS.Kanamalla,andVasileiosMegalooikonomou,ClassificationofbraintumorsusingMRIandMRSdata,Proc.SPIE6514,2007pp.65140S18.8A.A.Maudsley,C.Domenig,V.Govind,A.Darkazanli,C.Studholme,K.Arheart,C.Bloomer,MappingofbrainmetabolitedistributionsbyvolumetricprotonMRspectroscopicimagingMRSI,Magneticresonanceinmedicinevol.61,2009,pp.548559.9W.Dou,S.Ruan,Y.Chen,D.Bloyet,andJ.M.Constans,AframeworkoffuzzyinformationfusionforthesegmentationofbraintumortissuesonMRimagesImageandVisionComputing,vol.25,2007,pp.164–171.10WeibeiDOU,QianWU,YanpingCHEN,SuRUAN,andJeanMarcCONSTANS,FuzzymodellingofdifferenttumorouscerebraltissuesonMRIimagesbasedonfusionoffeatureinformation,Proceedingsof27thAnnualInternationalConferenceoftheIEEEEngineeringinMedicineandBiologySocietyEMBC2005,14September2005inShanghai,China.11WeibeiDOU,YuanREN,YanpingCHEN,SuRUAN,DanielBLOYET,andJeanMarcCONSTANS,HistogrambasedGenerationMethodofMembershipFunctionforExtractingFeaturesofBrainTissuesonMRIImages,LNAI2005Vol.3613,pp.189194.12LaraA.Brandao,RomeuC.Domingues,MRspectroscopyofthebrain,LivrariaeEditoraRevinterLtda.200313DenisHoa,Metabolitesexploredin1HMRS,http//www.imaios.com/en/eCourses/eMRI/MagneticResonanceSpectroscopyMRS14JeanMarcConstans,Variabilitysourcesinsinglevoxel1HMRSquantizationinbrain,sciencethesisofUniversitédeCaen/BasseNormandi,UFRdeMédecine,spécialitédeRecherchedinique,innovationtechnologie,santépublique,2006.15WeibeiDOU,ShuaiWANG,ShaowuLI,JeanMarcCONSTANS,AutomaticDataProcessingtoRelativeQuantitativeAnalysisof1HMRSpectroscopyofBrain,inproceedingsofThe3rdinternationalconferenceonbioinformaticsandbiomedicalengineeringiCBBE2009,June1116,2009,Beijing,China16WeibeiDou,SegmentationdimagesmultispectralesbaséesurlafusiondinformationsapplicationauximagesIRMPhD.Thesis,lUNIVERSITEdeCAEN,soutenule29septembre2006.

注意事项

本文(外文资料-- Glioma Tissue Modeling by Combing the Information of MRI and in vivo Multivoxel MRS.PDF)为本站会员(美女来了)主动上传,人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对上载内容本身不做任何修改或编辑。 若此文所含内容侵犯了您的版权或隐私,请立即通知人人文库网([email protected]),我们立即给予删除!

温馨提示:如果因为网速或其他原因下载失败请重新下载,重复下载不扣分。

copyright@ 2015-2017 人人文库网网站版权所有
苏ICP备12009002号-5