Classification of Alzheimer's Disease Integrating Individual Feature and Fusion Feature

Abstract

Intelligence diagnosis has been widely studied in the diagnosis of Alzheimer's disease (AD), but existing intelligent modeling methods cannot make full use of the multi-modal data feature information. As a result, the recognition accuracy is not high in the diagnosis of disease in the early stage. In order to improve the diagnosis effect of AD and its early stage, a classification method of Alzheimer's disease integrating individual feature and fusion feature is proposed. Firstly, the hypergraph convolutional network (HGCN) is used to extract features from the data of three modalities of MRI (magnetic resonance imaging), PET (positron emission computed tomography), and CSF (cerebro-spinal fluid) to obtain the high-order deep feature. At the same time, the data of these three modalities are fused through low-rank multimodal fusion to obtain hidden correlation features among multiple modalities. Finally, a multi-view classifier is used to comprehensively classify the above-obtained features. The ADNI dataset is used to classify AD in multiple groups of tasks to verify the proposed method. Compared with other state-of-the-art methods, the proposed method effectively improves the classification accuracy in the early stage of the disease while ensuring the classification effect of the AD stage.

Keywords

• multi-modal; hypergraph convolutional network (hgcn); low-rank multimodal fusion; multi-view classification; alzheimer's disease (ad)