Diffusion-guided MR Brain Tumor Segmentation with Missing Modalities

Motivation: Accurate brain tumor segmentation is crucial for effective diagnosis and treatment but is often complicated by missing MRI modalities in clinical practice。 Goal(s): This study introduces a diffusion-guided multi-modal segmentation framework designed to handle missing MRI modalities, a frequent challenge in clinical tumor segmentation. Approach: By combining a CNN-based model and a diffusion-based model, the framework adapts to incomplete data, providing robust and accurate segmentation results. Results: Testing on the BRATS2023GLIT dataset shows that this approach outperforms conventional methods, demonstrating improved segmentation. Impact: This approach enhances brain tumor segmentation accuracy and consistency, even with missing MRI modalities, thereby improving diagnostic precision and supporting clinical decision-making.