An Invertible Dynamic Graph Convolutional Network for Multi-Center ASD Classification

• DOI: 10.3389/fnins.2021.828512
• Publication Date: 2022-02-04T07:29:12Z
• Authors: Yueying Chen; Aiping Liu; Xueyang Fu; Jie Wen; Xun Chen
• Keywords: Interpretability; Discriminative model

Autism Spectrum Disorder (ASD) is one common developmental disorder with great variations in symptoms and severity, making the diagnosis of ASD a challenging task. Existing deep learning models using brain connectivity features to classify still suffer from degraded performance for multi-center data due limited feature representation ability insufficient interpretability. Given that Graph Convolutional Network (GCN) has demonstrated superiority discriminative representations networks, this paper, we propose an invertible dynamic GCN model identify investigate alterations patterns associated disease. In order select explainable model, blocks are introduced whole network, able reconstruct input network's output. A pre-screening adopted reduce redundancy information, fully-connected layer added perform classification. The experimental results on 867 subjects show our proposed method achieves superior disease classification performance. It provides interpretable analysis potential studying brain-related disorders.