See Prabhakar et al., MICCAI 2023
Multiple Sclerosis (MS) is a severe neurological disease characterized by inflammatory lesions in the central nervous system. Predicting inflammatory disease activity is crucial for disease assessment and treatment. In this work, the Menze group presents the first attempt to utilize graph neural networks (GNN) to aggregate MS biomarkers for a novel global representation. A two-stage MS inflammatory disease activity prediction approach is proposed that detects lesions using a 3D segmentation network and extracts their image features using a self-supervised algorithm. The detected lesions are used to build a patient graph, where the lesions act as nodes that are connected based on spatial proximity and the inflammatory disease activity prediction is formulated as a graph classification task. A self-pruning strategy auto-selects the most critical lesions for prediction. The proposed method outperforms the existing baseline by a large margin and offers inherent explainability by assigning an importance score to each lesion for the overall prediction.
Code is available at https://github.com/chinmay5/ms_ida.git.