GLI-AL: A Multi-Modal Glioma MRI Label Resource with Unified Anatomy-Lesion Labels
GLI-AL creates a unified, eight-class anatomy-lesion label set from BraTS data to address label noise from coexisting white matter hyperintensities (WMH). It achieves this by aligning lesion and healthy tissue into a single coding system, allowing machine learning models to train on joint supervision targets without switching between incompatible labels. This resource provides a WMH-aware reference for glioma segmentation research.