CoMNeT: A MedNeXt-CorrDiff Framework for Volumetric Brain Tumor Segmentation
Researchers have developed CoMNeT, a novel framework combining MedNeXt and CorrDiff for enhanced volumetric brain tumor segmentation from MRI scans. This approach utilizes four MRI modalities and incorporates a corrective diffusion model as a postprocessing step to refine segmentation accuracy. CoMNeT demonstrated superior performance on the UTSW-Glioma dataset compared to baseline models, achieving high Dice scores across different tumor regions. AI