Parameter Estimation in a Model Based Approach

《Parameter Estimation in a Model Based Approach》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。

1、2005IEEENuclearScienceSymposiumConferenceRecordM07-299ParameterEstimationinaModelBasedApproachforTomographicPerfusionMeasurementChristophNeukirchenandGeorgRoseAbstract—Thealgorithmicaspectsoftomographicestimationacquisitionandrelaxestherequirementforthelargenumberofofperfusionmodelparametersfr

2、omprojectionsacquiredwithaapproximatelyinstantaneousprojections.Inthenextsectionwerotatingimagingsystemarepresented.Theapproachisstatedasdescribeamethodfortheestimationofthemodelparametersamodelbasedparameteroptimizationproblem.Aparametricforthedynamicsinallimagepixelsfromnon-instantaneousmode

3、lisusedtodescribethetimeattenuationcurvesofthedynamicimageunderconsideration.Computedforwardpro-projections.InSec.IIIwepresentfirstresultsdeployingthejectionsofthismodelconsideringthetemporalandgeometricnewmethodbothonsimulationdataaswellasonstandardCTpropertiesoftheacquisitionsystemarecompared

4、tothemeasuredperfusiondatayieldingafirstproofofconcepts.projectionsinaLeastSquaresoptimization.AniterativeLanczosalgorithmisusedtosolvethisproblemnumerically.AnefficientimplementationispresentedwhichexploitsadecompositionofII.PARAMETERESTIMATIONMETHODtheproblemintothespatialandtemporaldomain.The

5、ill-posedThebasicproceduretoestimatethedynamicobjectgivennatureoftheproblemistackledbyaTikhonov-typeregularizationthenon-instantaneouslyacquiredprojectionsisbasedonthemethodthatmakesuseofaspatial/temporaldecompositionalongfollowingsteps:withadatadrivenoptimizationoftheregularizationparameter.T

6、hepotentialofthisapproachisshownfortwodifferentsetups:a1)Amodelthatdescribesthetemporalandspatialbrainperfusionsoftwarephantom,aswellasreconstructionsfromstructureofthedynamicobjectwhichissubjecttoastandardCTperfusionacquisitionofanacutestrokepatient.reconstructionmustbedefined.Suchmodeliscompl

7、etelyExperimentsshowthataccurateestimatesoftheunderlyingtime-determinedbyasetofparameterswhichhastobeintensitycurvescanberecoveredonapixel-by-pixelbase.estimatedfromtheacquiredprojections.IndexTerms—Perfusion,parameterestimation,tomogra