A Novel Nomogram for Predicting Meningioma Grade Based on Radiomics Features and Clinical Characteristics

OBJECTIVE: This study aimed to develop a predictive model utilizing radiomics features and clinical characteristics to accurately differentiate low-grade (WHO grade I) from high-grade (WHO grade II/III) meningiomas preoperatively, thereby improving treatment planning and prognosis. METHODS: A retrospective analysis of 288 meningioma cases (191 low-grade and 97 high-grade) confirmed by histopathology was conducted. Radiomics features were extracted from contrast-enhanced T1-weighted MRI (CE-T1WI) using the pyradiomics package, followed by feature selection via LASSO regression. Predictive models (logistic regression, decision tree, support vector machine SVM, adaptive boosting) were evaluated. Clinical variables (peritumoral edema index and monocyte count) were integrated to try to improve the predictive performance. Model efficacy was assessed using receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis. RESULTS: Four key radiomics features were identified as significant discriminators of tumor grade. The logistic regression model demonstrated superior predictive performance over decision trees, SVMs, and adaptive boosting methods. The inclusion of the peritumoral edema index and monocyte count increased the AUC to 0.801 (95% CI 0.753-0.869) in the training set. However, in the validation set, the radiomics model achieved the best performance, with an AUC of 0.770 (95% CI 0.670-0.869). CONCLUSIONS: The radiomics-based model effectively predicts high-grade meningioma and demonstrates superior performance compared to the clinical and combined models. This study advances the precision of meningioma grading, offering significant implications for treatment planning and patient management.