evolvable-graph-diffusion-ot

Align structural and functional brain connectomes by inputting SC and FC matrices into the model, which uses a learnable optimal transport mechanism to drive accurate disease classification predictions.

Evolvable graph diffusion is a mechanism that dynamically captures higher-order dependencies across brain regions, moving beyond fixed structural-functional alignments to improve multi-modal brain network analysis.

Yes, you can use this PyTorch-based model for brain disease classification by inputting your structural and functional connectome matrices to obtain predictions for conditions like MDD or AD.

KAN node aggregation uses neural spline-based transformations to provide flexible, non-linear feature integration, enabling more accurate structure-function relationship studies in neuroimaging workflows.

Fixed SC-FC alignments lack adaptability, whereas learnable optimal transport dynamically learns a transport plan to align structural and functional networks, addressing limitations of static approaches.

Yes, you need pre-computed structural connectivity and functional connectivity brain connectome matrices to input into the model to perform end-to-end optimal transport alignment and classification.