dgcl-brain-network-construction

You can construct a brain network from fMRI data by using diffusion-based brain region-aware modules and graph contrastive learning to automate the end-to-end pipeline, enabling disease-stage prediction for disorders like Alzheimer's and Autism.

Graph contrastive learning is used in brain network construction to improve consistency and reduce manual parameter tuning by jointly optimizing node and graph representations, which helps identify disease-related connections from fMRI data.

Yes, this approach supports Alzheimer's disease-stage prediction using the ADNI dataset, applying diffusion-based localization and graph contrastive learning to reconstruct brain networks and extract key disease-related connections.

The diffusion-based brain region-aware module works by localizing brain regions in fMRI data to construct an initial network, which is then refined through graph contrastive learning and joint loss optimization for accurate disease analysis.

No, you do not need to manually tune parameters extensively; graph contrastive learning and joint node-graph optimization reduce manual tuning, ensuring efficient and reproducible workflows for brain disorder analysis.

Yes, this method supports Autism dataset analysis using the ABIDE dataset, allowing you to perform disease-stage prediction and identify important disease-related connections through an end-to-end fMRI pipeline.