Abstract
PURPOSES: The preoperative distinction between atypical meningioma (AM) and intracranial solitary fibrous tumor (SFT) holds significant importance in guiding surgical approach decisions and prognostic assessments. METHODS: A total of 310 SFT patients and 203 AM patients were retrospectively included and stratified into training and validation cohorts. Employing the elastic net algorithm, relevant features were identified to form the fusion radiomic model. Subsequently, a clinical-radiomic combined model was developed by integrating the fusion radiomic model with significant clinical variables through multivariate logistic regression analysis. The models' calibration, discriminative capacity, and clinical utility were thoroughly assessed. RESULTS: The fusion radiomic model was crafted from 17 radiomic features, achieving AUC values of 0.920 in the training set and 0.870 in the validation set. Subsequently, the clinical-radiomic combined model exhibited AUC values of 0.930 and 0.890 in the training and validation sets, indicating commendable discrimination and calibration. Assessment through decision curve analysis underscored the clinical utility of both the fusion radiomic model and the clinical-radiomic combined model for individuals with intracranial SFT and AM. CONCLUSIONS: The clinical-radiomic combined model exhibited notable sensitivity and exceptional efficacy in the distinctive diagnosis of intracranial SFT and AM, holding promise for the non-invasive advancement of personalized diagnostic and therapeutic strategies.