A research team involving the DQBM Menze Lab has introduced GlioMODA, a deep learning framework designed to automate glioma segmentation in routine clinical imaging. The study evaluated the method across 11 MRI sequence combinations using 1,470 cases from the BraTS 2021 dataset.
GlioMODA achieved segmentation accuracy comparable to a standard four-sequence protocol when using only contrast-enhanced T1 and T2-FLAIR images. The framework could support more efficient clinical workflows and reproducible quantitative tumour assessment when complete MRI protocols are unavailable. The authors have released the models and code as an open-source package.