基于超图的阿尔茨海默病辅助诊断研究现状与进展

Research work and progress of aided diagnosis of Alzheimer’s disease based on hyper graph

阿尔茨海默病(Alzheimer's Disease,AD)等神经变性疾病普遍存在,记忆危机日益严重,因此对于早期AD辅助诊断的需求十分迫切。医学影像技术是前期辅助AD筛查的一种有效手段,其中超图在对AD分类任务中表现突出。在超图中,超边可以连接多个节点,这使得超图更适用于表示复杂的关系和结构。在医学影像技术中,超图能够更加准确地建模多元关系,具备较强的数据样本间非线性高阶关联的刻画和挖掘能力。汇总了基于脑功能超网络的研究成果,重点介绍了图核、矩阵分析、深度学习这3种方法,最后对未来的发展进行展望,为后续研究提供参考。

With the prevalence of neurodegenerative diseases such as Alzheimer's disease(AD)and the growing memory crisis,there is a significant need for early AD based diagnostic aids.Medical imaging technology is an effective tool to assist in AD screening in the early stages,where hypergraphs excel in the task of classifying AD.In hypergraphs,hyperedges can connect multiple nodes,which makes hypergraphs more suitable for representing complex relationships and structures.In medical imaging technology,hypergraphs are able to model multivariate relationships more accurately,with a strong ability to portray and mine nonlinear higher-order associations between data samples.This paper summarises the research results based on brain functional hypernetworks,focusing on three methods:graph kernels,matrix analysis and deep learning,and concludes with an outlook on future developments to provide a reference for subsequent research.