脑白质疏松症MR图像病变区域的量化分析  

Lesion Areas Quantitative Analysis for Leukoaraiosis MR Image

作　　者：赵波[1] 郑兴华[1] 白璐[1] 杨勇[2] 

机构地区：[1]杭州电子科技大学自动化学院,浙江杭州310018 [2]杭州电子科技大学生命信息与仪器工程学院,浙江杭州310018

出　　处：《杭州电子科技大学学报（自然科学版）》2012年第6期117-120,共4页Journal of Hangzhou Dianzi University：Natural Sciences

基　　金：国家自然科学基金资助项目(30770685);浙江省重大科技专项(优先主题)国际科技合作资助项目(C14078)

摘　　要：该文提出了一种基于图像配准和分割的方法来对脑白质疏松症MR图像中脑白质病变区域进行量化。首先,利用标记分水岭法在T1flair序列上将头骨等组织去除,并利用多尺度小波变换法提取出脑白质区域;其次,通过对T1和T2flair序列进行配准,提取出T2flair序列上脑白质区域,并采用Otsu阈值法,在相应的T2序列脑白质区域图上分割出脑白质病变区域;最后,通过分割后计算得到的脑白质体积值和病变区域的体积值,计算出病变区域所占百分比。结果表明,该方法实现了对脑白质疏松症MR图像的计算机自动分割,能够衡量脑白质疏松症的病变程度。因此,此方法在脑白质疏松症辅助诊断上具有临床应用价值。To analyze quantitatively of the lesion areas in leukoaraiosis MR image,a method of registration and segmentation was presented in this paper.First,the skull portion of MR image on T1 flair sequence was removed by marker-driven watershed algorithm,then multi-scale wavelet transform method was applied to extract the white matter regions;Second,the white matter regions on T2 flair sequence would be extracted by image registration of T1 T2 flair,and then by using Otsu threshold method,the white matter lesions would be separated from the corresponding T2 white matter area;Finally,the volume percentage of white matter lesions would be calculated from the segmented white matter volume and white matter lesion volume.Results show this method can realize leukoaraiosis＇ MR image segmentation automatically,which can further evaluate the severity of leukoaraiosis.Therefore,this method has clinical value on assistant diagnosis to leukoaraiosis.

关 键 词：脑白质疏松症 磁共振图像 图像配准 图像分割 

分 类 号：R391[医药卫生—基础医学]