论文部分内容阅读
为了定位颞叶癫痫(TLE)患者脑白质微结构发生异常的重要脑区,本文设立了正常对照组(NC)与TLE组两组人群,采集了50位受试者(其中NC组28人,TLE组22人)的脑部弥散张量成像(DTI)影像,分别计算其部分各向异性(FA)、平均扩散率(MD)、扩散系数(AD)、径向扩散系数(RD)等参数,并采用纤维束追踪空间统计方法(TBSS),获取组间差异的脑区,然后利用支持向量机(SVM),对NC组与TLE组进行分类,并与支持向量机-递归特征消除法(SVM-RFE)进行比较,最后对重要脑区及其分布进行分析与讨论。实验结果表明,TLE患者的FA值存在明显降低的脑区主要有胼胝体、上纵束、放射冠、外囊、内囊、下额枕束、钩束、矢状层等,基本呈双侧分布,其中大部分脑区的MD、RD值明显增高,AD值虽有增高,但差异无统计学意义。支持向量机-纤维束追踪空间统计法(SVM-TBSS)利用FA、MD、RD进行分类的准确率分别为82%、76%、76%,特征融合后分类准确率为80%;SVM-RFE利用FA、MD、RD进行分类准确率分别为90%、90%和92%,特征融合后分类准确率达到100%,SVM-RFE分类性能明显优于SVM-TBSS,对分类有重要影响的特征主要分布于联络纤维和连合纤维脑区。研究结果表明,DTI参数能有效地反映TLE患者的脑白质纤维异常改变,可用于阐明其病理机制、定位病灶及实现自动诊断。
In order to locate an important cerebral area of abnormal white matter structure in patients with temporal lobe epilepsy (TLE), two groups of normal control group (NC) and TLE group were established. Fifty subjects (28 in NC group, (TLE group, 22 persons). The parameters of partial anisotropy (FA), average diffusivity (MD), diffusion coefficient (AD) and radial diffusion coefficient (RD) (TBSS) was used to obtain the brain regions of difference between the groups. Then the SVM and the SVM were used to classify the NC group and the TLE group, and compared with the support vector machine - recursive feature elimination method SVM-RFE), and finally analyze and discuss the important brain regions and their distribution. The experimental results showed that there were mainly corpus callosum, upper longitudinal bundle, coronal coronaries, outer capsule, inner capsule, inferior frontal occipital bundles, hook bundle and sagittal layer in FA patients with TLE. , Most of which MD, RD value was significantly higher, although the AD value increased, but the difference was not statistically significant. SVM-TBSS classification accuracy by using FA, MD and RD were 82%, 76% and 76%, respectively, and the classification accuracy after feature fusion was 80%. SVM-RFE The classification accuracy of FA, MD and RD were 90%, 90% and 92% respectively, and the classification accuracy of the feature fusion was 100%. The classification performance of SVM-RFE was superior to that of SVM-TBSS. Mainly distributed in the liaison fibers and commissural fiber brain area. The results show that DTI parameters can effectively reflect the abnormal changes of white matter in patients with TLE, which can be used to clarify the pathological mechanism, locate the lesion and achieve automatic diagnosis.