【摘要】 针对流行的磁共振扩散张量成像数据分析工具TBSS,提出了一种定量评估的方法。首先构建模拟数据,由31个正常人的DTI-FA图像人工模拟得到多组白质不同程度损伤程度的阿兹海默症病人组数据。将模拟得到的多组病人数据分别与正常组进行基于体素的TBSS分析。通过计算分析结果中有显著性差异的体素数量和真实变化体素数量之间的比例关系,得到该分析方法的敏感性和特异性指标。结果显示TBSS分析敏感性增大的同时伴随着特异性的减小。当模拟病人组FA值减小幅度比较大时,敏感性可达到90%以上,但特异性会不足50%,甚至更低。因此在使用TBSS进行相关疾病研究时,分析结果的判读要考虑到可能的假阳性结果,应结合其他的医学实验和研究结果进行综合分析。更多还原

【Abstract】 An evaluative pipeline was developed to quantitatively assess TBSS,which is widely used in voxelbased diffusion tensor imaging analysis. First,datasets of Alzheimer’s disease patient group with white matter injury of different degrees were simulated from the DTI-FA images of 31 normal controls. Then the simulated datasets were analyzed with TBSS. The numbers of the detected voxels changed significantly and the ground-truth changed voxels were obtained. The sensitivity and specificity of the method were thus calculated. Quantitative results showed that the sensitivity of TBSS analysis was increased at the cost of the decrease of specificity. When the decreased FA values in patient group was relatively large,the sensitivity could be up to 90%,while the specificity would decrease to 50%,or even lower. Therefore,when interpreting the TBSS analysis results,other complementary medical studies should be considered to account for the relatively higher false positive rate.更多还原