基于改进Marching Cubes算法的乳腺MRI肿块三维重建.pdf

《基于改进Marching Cubes算法的乳腺MRI肿块三维重建.pdf》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、第26卷第4期传感技术学报Vo1．26No．42013年4月CHINESEJOURNALOFSENSORSANDACTUATORSApt．2013Three-DimensionalRec0nstructi0n0fBreastMRIMassesBasedonImprovedMarchingCubesAlgorithmZHUYimiao，XUWeidong，LILihua，LIUWei，XU，ZHANGJuan(1．CollegeLifeInformationScience＆InstrumentEngineerin

2、g，HangzhouDianziUniversity，Hangzhou310018，China；2．DepartmentRadiology，ZhefiangCancerHospita，Hangzhou310022，China)Abstract：MRI(MagneticResonanceImaging)hasbecomeoneoftheprimarymeansofbreastcancer’Sdiagnosis．AnimprovedmethodbasedonMC(MarchingCubes)surfacerende

3、ringalgorithmisproposed，whichisappliedtothethree-dimensionalreconstructionofthebreastMRImasses．ByanalyzingtheimagingcharacteristicsofbreastMRIsequencesandthecorrelationbetweenadjacentlayers，whichincludethesimilarityofadjacentinterframeimagegraydistributionan

4、dmassorganizationsshape，anisosurfaceextractionalgorithmbasedonRSF(Region-scalableFitting)isproposedfirstly．Then，inaccordancewiththeconditionsoftheoverlappingareaofeachframewithitsfrontandrearflamethebiggerthebetter，filtertheresultsobtainedunderthemultipleset

5、sofparameterstooptimizetheextractionofcontoursurface．Finally，simplifyalargenumberoftriangularmeshgeneratedinsurfacerendering，byusingthetriangularfoldingmethodbasedonweightedquadraticerrormetric．Theproposedmethodwasappliedto30casesofbreastMRIsequencediagraman

6、dtheexperimentalresultsshowedthatitperformedbetterthantheconventionalMCalgorithmontheaccuracyandspeedofrenderingforthree-dimensionalreconstructionofthebreastMRImasses．Keywords：medicalimageprocessing；MRImass3Dreconstruction；inter—flamecorrelation；marchingcube

7、s；triangularmeshsimplificationEEACC：7510doi：10．3969／j．iSSli．1004—1699．2013．04．001基于改进MarchingCubes算法的乳腺MRI肿块三维重建木朱益苗，徐伟栋，厉力华，刘伟，徐平，张娟(1．生命信息与仪器工程学院，杭州电子科技大学，杭州310018；2．浙江省肿瘤医院放射科，杭州310022)摘要：核磁共振成像MRI(MagneticResonanceImaging)是目前乳腺癌肿块诊断的常用辅助手段，对图像的正确解析是关键，针对

8、传统MC(MarchingCubes)面绘制算法应用于乳腺MRI图像的不足，提出了改进方法。首先利用乳腺MRI序列图相邻帧间图像灰度分布的相似，肿块组织形状相近等相关性，在RSF(Region—ScalableFitting)模型的基础上利用初始轮廓迭代的方法提取肿块区域。接着将多组参数下获得的结果，依据每一帧与其前后帧的重叠面积越大越好作为条件进行筛选，使提取的等值面最优化。最后采用