Background <p>IDH and TERTp mutations serve as key molecular markers in glioma, closely associated with tumor clinical behavior, treatment response, and patient survival. Precise identification of molecular subtypes facilitates improved clinical risk stratification. This study aims to develop an integrated radiomics and pathomics model for the rapid classification of glioma molecular subtypes.</p> Methods <p>This multi-center retrospective study enrolled 434 adult patients with diffuse glioma from three independent centers. Radiomics features were extracted from preoperative MRI sequences, and pathomics features were extracted from regions of interest delineated on whole-slide images. Following feature selection, five distinct machine learning algorithms were employed to build individual radiomics and pathomics models for the binary classification of IDH and TERT promoter (TERTp) status, as well as for a four-class molecular subtype classification. The best-performing algorithm for each task was subsequently selected to construct a stacking model. Model classification performance was evaluated using receiver operating characteristic (ROC) curves and the area under the curve (AUC). </p> Results <p> For the binary classification of IDH mutation status, the stacking model integrating radiomics and pathomics achieved AUCs of 0.850 in validation set and 0.852 in external test set. For TERTp mutation status prediction, the stacking model yielded AUCs of 0.766 (validation) and 0.770 (external test). In predicting the four-class molecular subtypes combining IDH and TERTp status, the stacking model demonstrated micro-AUC/macro-AUC values of 0.835/0.824 in validation set and 0.827/0.815 in external test set, outperforming the predictive performance of the TERTp-only model.</p> Conclusions <p> Stacking model based on Radiopathomics can rapidly predict multi-class molecular subtypes of glioma, providing robust support for risk stratification and management in patients.</p>