Finite Element Modeling ofHumanHeadfrom Medical Images

Thefinite element (FE)modeling ofarealistic head TheCSF,GM andWM fromtheMRIimageandthescalp isa keyissue forthefinite element analysis ofbrain andskull fromtheCT imagemaynotbeinthesame electromagnetic field. Inthepresent study, wehavedeveloped a newmethodtogenerate subject specific FEheadmodels based cdif ntebeause thea Cimessare cquredith ontheir magnetic resonance (MR)andcomputer tomographydifferent imaging systems. ItiSthusnecessary toco-register (CT)imaging data. Thepresent approach consists ofthree parts: theMR andCTimages toputallthefive tissues into thesame segmentation ofMR andCTimages, co-registration ofMR and coordinate system. A co-registration algorithm(3) was CTimages, andmeshgeneration foralltissue volumes. Byusing developed inthepresent study basedonthecontour ofthe thismethod, afive-layer realistic headFEM modelisobtained,images, whichincludes foursteps: (a)alignment ofthe andshowntoprovide reasonable EEG fields producedby corresponding slices inCTandMRI,(b)imagedescription by current dipoles. randomsignal sequences, (c)objective function construction I. INTRODUCTION basedon randomsignal sequences, and (d)space transformation matrix construction. Finite element method(FEM) hasbeenshowntobeuseful Fig. 1showstheresult ofsegmentation andco-registration . inanalyzing EEGinaninhomogeneous headmodel. However, itisstill challenging toconstruct aFEM headmodelfrom subjects' MRIorCTimages. Inthepresent study, wedevelop an automatic methodtogenerate theFEM modelofan individual subject. First, we obtain asharply defined image of eachtissue fromthesamesubject's MRI andCT databy image segmentation. Then, through image co-registration, we getthefused image that contains theinformation inMRIand CT.Thesegmented andco-registered tissues arethenmeshed(a b(c into wellshaped finite element units andtheFEM modelof