颞叶癫痫患者脑白质纤维束追踪空间统计分析与自动识别

发布时间：2018-03-14 05:53

本文选题：颞叶癫痫　切入点：弥散张量成像　出处：《生物医学工程学杂志》2017年04期 　论文类型：期刊论文

【摘要】：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。
[Abstract]:In order to locate the important areas of white matter microstructures in patients with temporal lobe epilepsy (Tle), two groups of normal control group and TLE group were established. DTI images were collected from 50 subjects (including 28 patients in NC group and 22 in TLE group). The parameters such as partial anisotropy, average diffusion rate (MDN), diffusion coefficient (ADT) and radial diffusion coefficient (RD) were calculated, respectively. The spatial statistical method of traceability of fiber bundle was used to obtain the different brain regions between groups. Then, support vector machine (SVM) was used to classify NC group and TLE group, and compared with SVM-RFE (support vector machine-recursive feature elimination method). Finally, the important brain regions and their distribution were analyzed and discussed. The results showed that the main areas in which the FA values of TLE patients were significantly decreased were corpus callosum, superior longitudinal tract, coronal radiate, external capsule, internal capsule, inferior frontal occipital bundle, hook bundle, sagittal layer, etc. The MDR value in most of the brain regions was significantly increased, although the AD value was increased. But the difference was not statistically significant. The accuracy of SVM-TBSS-SVM-TBSS-SVM-TBSS-based classification was 822276 / 767, respectively, and that of SVM-RFE after feature fusion was 90% and 922% respectively, and the classification accuracy of SVM-RFE was 90% and 922%, respectively, after the feature fusion, the classification accuracy of SVM-TBSSS was 90% and 922% respectively, and the classification accuracy of SVM-RFE was 90% and 922% respectively. The classification accuracy of SVM-RFE is 100% after feature fusion, and the classification performance of SVM-RFE is obviously better than that of SVM-TBSS.The main features that have important influence on classification are distributed in the brain regions of contact fibers and conjunctive fibers. The results show that the parameters of SVM-RFE can effectively reflect the brain of TLE patients. Abnormal changes in white matter fibers, It can be used to clarify the pathological mechanism, locate the lesion and realize automatic diagnosis.
【作者单位】：华南理工大学材料科学与工程学院生物医学工程系;广东三九脑科医院影像诊断中心;
【基金】：国家自然科学基金资助项目(31371008) 广东省科技计划支撑项目(2015A02024006) 广州市产学研协同创新重大专项(201604020170)
【分类号】：R742.1;TP18

【相似文献】