multimodal-neuroimaging

Artemis proposes a region-level causal framework that learns region-specific confounder representations to eliminate demographic confounders in multimodal neuroimaging, improving graph neural network performance on disease diagnosis and classification tasks.