Deep Structure Learning on Brain Networks for FMRI Analysis

Functional magnetic resonance imaging (fMRI) has become one of the most common imaging modalities for brain function analysis. Recently, graph neural networks (GNNs) have been adopted for fMRI analysis with superior performance. Unfortunately, traditional functional brain networks are mainly constructed based on similarities among regions of interests (ROI), which are noisy and can lead to inferior results for GNN models. To better adapt GNNs for fMRI analysis, we propose DSBNet, a \textbf{D}eep \textbf{S}tructure learning framework based on \textbf{B}rain \textbf{Net}works for fMRI analysis. DSBNet adopts a brain network generator module, which harnesses the DAG learning approach to transform the raw time-series into effective brain connectivities. Experiments on two fMRI datasets demonstrate the efficacy of DSBNet. The generated brain networks also highlight the prediction-related brain regions and thus provide interpretations for predictions.