Overview
BraTS research uses multimodal brain MRI to identify and characterize tumors, especially by segmenting whole tumor, tumor core, and enhancing tumor regions. These regions support clinical assessment and provide a shared basis for comparing methods, while related work explores tumor grading, infiltration risk, MRI synthesis, and inpainting. Common approaches include 3D U-Net-style networks, transformers and state-space models, modality-fusion methods, and techniques for missing inputs, limited labels, or image degradation. A central challenge is maintaining reliable performance across hospitals, cohorts, and imaging conditions, with evaluation complicated by differences in data splits, metrics, and whether models use clinically appropriate evidence.
Latest papers
29 papers4 papers in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.