范文亮¹  李灵丽¹   杨炼¹   朱朝辉²

（１华中科技大学同济医学院附属协和医院放射科，２华中科技大学同济医学院附属协和医院泌尿外科）

摘要：目的: 探讨静脉型勃起功能障碍大脑结构改变及其与临床症状的关系，进一步通过机器学习分类区分病例组与对照组。方法：收集45例静脉型勃起功能障碍患者和50例对照组。采用基于体素形态学方法、基于白质骨架的空间统计方法和相关性分析。机器学习分类进一步证实其区分病例组与对照组的实效性。结果：与对照组相比，病例组左侧中央后回和中央前回皮质体积减少，但右侧颞中回的皮质体积增加；广泛脑区的轴向扩散系数、横向扩散系数和平均扩散系数增加，而且，病例组一些脑区的改变与临床症状显著相关。机器学习区分病例组与对照组的准确度96.7%、灵敏度93.3%和特异度99.0%。结论：静脉型勃起功能障碍显示皮质体积和白质纤维结构改变，这些改变与临床症状显著相关。而且，机器学习证实一些脑区的扩散张量相关标量指标可作为病例组与对照组可靠的识别特征。

Abstract：Objectives: To investigate cerebral structural changes related to venous erectile dysfunction(VED) and relationship with their clinical symptoms, and to distinguish patients with VED from healthy controls by machine-learning classification.Methods: A total of 45 VED patients and 50 healthy controls were included. VBM, TBSS and correlation analysis were performed. Machine-learning classification was adopted to confirm the effectiveness in distinguishing VED versus healthy controls.Results: Compared to healthy control subjects, the VED patients showed signifcant decreased cortical volumes in left postcentral gyrus and precentral gyrus, while only right middle temporal gyrus showed significant increased. Increased axial diffusivity(AD), radial diffusivity (RD) and mean diffusivity(MD) values was observed in widely brain regions. Moreover, certain regions of these alterations related to VED patients showed a significant correlation with clinical symptoms. Machine-learning analyses discriminated the patients from controls with an overall accuracy of 96.7%, sensitivity of 93.3% and specificity of 99.0%.Conclusions: The cortical volumes and WM microstructural changes were observed in VED patients, and these alterations showed a significant correlation with clinical symptoms. Furthermore, the multiple DTI-derived indices of some brain regions could be regarded as reliably discriminating features between VED patients and healthy control subjects by machine-learning analyses.