基于几何深度学习的脑形态学研究在阿尔茨海默病诊断中的初步应用  

A Preliminary Study of Applying Geometric Deep Learning in Brain Morphometry for Diagnosis of Alzheimer’s Disease

作　　者：谢薇[1] 孙怀强[2,3] 陈嘉伟 曾意 徐旭 李真林 夏春潮[1] XIE Wei;SUN Huai-qiang;CHEN Jia-wei;ZENG Yi;XU Xu;LI Zhen-lin;XIA Chun-chao(Department of Radiology,West China Hospital Sichuan University,Chengdu 610041,China;Huaxi MR Research Center,Department of Radiology,West China Hospital,Sichuan University,Chengdu 610041,China;Functional and Molecular Imaging Key Laboratory of Sichuan Province,Chengdu 610041,China;Department of Ultrasound Medicine,West China Hospital,Sichuan University,Chengdu 610041,China)

机构地区：[1]四川大学华西医院放射科,成都610041 [2]四川大学华西医院放射科临床磁共振研究中心,成都610041 [3]功能与分子影像四川省重点实验室,成都610041 [4]四川大学华西医院超声医学科,成都610041

出　　处：《四川大学学报（医学版）》2021年第2期300-305,共6页Journal of Sichuan University(Medical Sciences)

基　　金：国家自然科学基金(No.81974278);中国科协青年人才托举工程资助项目(No.YESS20160060);四川大学华西医院学科卓越发展1·3·5工程项目(No.ZYGD18019)资助。

摘　　要：目的基于脑表面图形和几何深度学习建立阿尔茨海默病(Alzheimer’s disease,AD)的分类预测模型,并评估其性能。方法纳入临床确诊AD患者76例,健康老年人83例,并按4∶1的比例随机划分为训练集和测试集。从受试者的MR成像中三维T1加权高分辨率结构像中构建脑表面图形,进行一系列图形简化操作后将训练集输入几何深度神经网络进行训练,用测试集对训练产生的预测模型进行性能评估,评估参数包括准确率、敏感性和特异性。结果在右脑面数为6 000的脑表面图形上训练得到的预测模型取得最佳性能(准确性93.8%,敏感性91.7%,特异性94.1%)。脑表面图形在卷积与池化操作过程中的变化揭示AD患者相较健康老年人存在全脑弥漫分布的脑组织损失。结论基于图形数据和几何深度学习的脑形态学分析方法在AD的诊断和鉴别诊断中有较大的发展潜力。Objective A predictive model of Alzheimer’s disease(AD) was established based on brain surface meshes and geometric deep learning, and its performance was evaluated. Methods Seventy-six clinically diagnosed AD patients and 83 healthy older adults were enrolled and randomly assigned to the training set and the test set according to a4-to-1 ratio. Brain surface mesh was constructed from 3-D T1-weighted high-resolution structural MR volumes of each participant. After applying a series of simplification to the surface meshes, the training set was fed into the geometric deep neural network for training. The performance of the prediction model was evaluated with the test set, and the evaluation metrics included accuracy, sensitivity and specificity. Results The prediction model trained on the right brain surface meshes with 6 000 faces achieved the best performance, with accuracy reaching 93.8%, sensitivity, 91.7%, and specificity,94.1%. The evolution of the brain surface meshes during convolution and pooling revealed that AD patients had diffuse brain tissue loss compared with healthy older adults. Conclusion Morphological brain analysis based on mesh data and geometric deep learning has great potential in the differential diagnosis of AD.

关 键 词：阿尔茨海默病 脑形态学 磁共振脑影像 几何深度学习 

分 类 号：R749.16[医药卫生—神经病学与精神病学]