Repo with code associated with paper "Multimodal Graph Coarsening for Interpretable, MRI-Based Graph Neural Network"
MMSAG.py and MMTopK.py provide implementations of the multimodal SAGPooling and TopK pooling layers described in the paper for M=2 input modalities.
MMGNN.py provides code for the model architecture used to obtain the results on the COBRE case/control dataset, accessible here.
The model can be integrated into a standard PyTorch training script using data formatted with the PairData function provided by PyTorch Geometric.