1 B-spline registration of 3D images with Levenberg-Marquardt optimization

1 B-spline registration of 3D images with Levenberg-Marquardt optimization

ID:39459122

大小:301.50 KB

页数:10页

时间:2019-07-03

1  B-spline registration of 3D images with Levenberg-Marquardt optimization_第1页
1  B-spline registration of 3D images with Levenberg-Marquardt optimization_第2页
1  B-spline registration of 3D images with Levenberg-Marquardt optimization_第3页
1  B-spline registration of 3D images with Levenberg-Marquardt optimization_第4页
1  B-spline registration of 3D images with Levenberg-Marquardt optimization_第5页
资源描述:

《1 B-spline registration of 3D images with Levenberg-Marquardt optimization》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、B-SplineRegistrationof3DImageswithLevenberg-MarquardtOptimizationSvenKabusa,b,ThomasNetschb,BerndFischera,JanModersitzkiaaInstituteofMathematics,UniversityofLubÄeck,Wallstra¼e40,23560LubÄeck,GermanybPhilipsResearchLaboratories,RÄontgenstra¼e24,22335Hamburg,GermanyABSTRACTB-splinesareawell-knownapp

2、roachfornon-rigidimageregistration.Thoughsuccessfullyappliedtovariousmedicalapplicationstheyexhibitahighcomputationalcomplexitymainlybecauseofthelackofdedicatedoptimizationmethods.InthisworkwefocusonaLevenberg-Marquardttypeoptimizationroutine.Asasimilaritymeasureweuseleast-squaresfunctionalssuchas

3、thesumofsquareddi®erences,thecross-correlationandthelocalcorrelationmeasure,respectively.Thelatterisusedformulti-modalityregistrationtasks.Theproposedregistrationalgorithmconsistsofthreemainparts.Ineachiterationsteponehasto(a)buildalinearsystemofequations,(b)solvethissystemandcomputeanupdate,(c)de

4、terminethesteplengthforthefollowingiterationstep.Appropriatestoppingcriteriaensuretheterminationoftheregistrationtask.Astandardapproachfor(c)andseveralmodi¯cationsareinvestigated.Usingaquadraticmodelweareabletoavoidadditionalexecutionof(b)duringthesteplengthadaption.Severalsolvers(Cholesky,CG,pre-

5、conditioning)for(b)havebeenevaluated.Also,modi¯cationsonthemosttimeconsumingtask(a)areinvestigated,leadingtoaspeed-upbyafactorupto30.Finally,thealgorithmisembeddedinamulti-scaleframework(bothonimageandonparameterlevel)providingadditionalregularization,anincreasedcapturerangeandspeed-up.Convergence

6、testshavebeensuccessfullyappliedforaprioriknowntransformations.FeasibilityoftheproposedapproachisalsoshownforclinicalapplicationsincludingPET-CTregistrations(19datasets)andMRmammography.Keywords:3Dimageregistration,B-splines,Levenberg-Marquardtoptimization,multi-scaledecomposition1.INTRODUCTIONThe

7、taskofnonrigidimageregistrationofthree-dimensional(3D)medicalimagevolumesisto¯ndavector¯eldof3Ddisplacementssuchthateachpointinthereferenceimagecanbemappedontoacorresponding(meaning-ful)pointinthetemplateimage.Ty

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。