Multicenter MRI Model with Habitat and Radiomics Predicts Glioma IDH Mutation

A combined MRI radiomics-habitat-clinical model predicted IDH mutation in gliomas with AUC 0.952 in training and 0.867-0.922 across three external test sets, with habitat-derived features being most influential.

• Multicenter retrospective study included 1076 glioma patients from four datasets with external validation in three independent sets.
• Combined model AUC: 0.952 (training), 0.867–0.922 (external); the habitat-score was the most important predictor by SHAP analysis.
• Transcriptomic analysis linked imaging signatures to IDH-related differences in proliferation, metabolism, extracellular matrix remodeling, and immune pathways.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.