基于动态图论特征的阿尔茨海默病早期预测

Early prediction of Alzheimer disease based on dynamic graph theory

目的阿尔茨海默病是常见的神经退行性疾病,而主观认知下降可发生于阿尔茨海默病临床前期,但传统磁共振数据分析对于主观认知下降的区分度较低。本研究通过动态图论特征在磁共振数据中的应用,研究主观认知下降患者脑动态功能连接的变化。方法随机纳入30例主观认知下降患者以及年龄、性别等相匹配的30例健康受试者的静息态磁共振数据,分别计算两组受试者时域变换灵活性和空间分布广泛性两个动态图论特征,同时构建基于支持向量机的分类器,研究动态图论特征的分类效果,分析两组受试者动态图论特征的主要分类贡献脑区的变化差异。结果研究发现主观认知下降患者在额上回、楔前叶等认知相关脑区的动态图论特征较之于正常受试者有显著减弱,且动态图论特征在分类中能够获得较高的准确度,优于传统静态图论特征。结论动态图论特征能够分析主观认知下降患者脑动态功能连接的变化,为阿尔茨海默病的早期诊断提供理论依据。

Objective Alzheimer disease is the most common neurodegenerative disease and subjective cognitive decline(SCD)is the early stage of Alzheimer disease.However,the recognition for SCD of traditional MRI analysis is low.This paper aims to find the changes of dynamic functional connectivity in SCD patients.Methods Temporal flexibility and spatiotemporal diversity of 30 individuals with SCD and 30 matched normal controls were calculated by using the resting-state functional MRI data.Classifiers based on support vector machine were constructed to study the classification effect of dynamic graph theory parameters.The differences of main brain regions chose by dynamic graph theory parameters were analyzed.Results We found that the dynamic graph features of cognitive-related brain regions such as frontal gyrus and anterior wedge of SCD patients were significantly weaker than normal subjects,demonstrated higher classification accuracies than conventional static parameters.Conclusions Dynamic graph theory features can reveal the changes of brain dynamic function connections of SCD patients,providing a theoretical basis for the early diagnosis of Alzheimer disease.