Classification of ADHD based on fMRI data with Graph Convolutional Networks

Attention deficit/Hyperactivity disorder (ADHD) is a neurodevelopmental disease diagnosed primarily by clinical scales, which is susceptible to individual subjectivity. In the classification of ADHD, traditional convolutional neural networks are not suitable for non-Euclidian spatial brain networks. Therefore, in this study, we focus on graph neural network to classify ADHD based on functional magnetic resonance imaging (fMRI) data. Each participant is treated as a node, the connectivity between nodes as an edge, and a corresponding brain functional connectivity network is assigned to each node. The results show that our method achieves 75% accuracy and an average AUC of 0.87, reaching a high level of classification for ADHD.