A deep learning-based radiomic nomogram derived from visceral fat for early prediction of gastrointestinal stromal tumor risk grade

Gastrointestinal stromal tumors (GISTs) are heterogeneous tumors classified into different risk categories requiring distinct treatment strategies. The study developed a deep learning-based radiomics nomogram (DLRN) using visceral fat features from non-contrast CT scans to predict GIST risk grade before surgery. The research included 211 patients with histologically confirmed GISTs from two institutions. In the derivation cohort, the DLRN achieved an area under the curve (AUC) of 0.936 with accuracy of 0.873, sensitivity of 0.811, and specificity of 0.904. In the external test cohort, the AUC was 0.862 with accuracy of 0.925, sensitivity of 0.833, and specificity of 0.936. The DLRN demonstrated higher clinical benefit compared to other models. The proposed nomogram may serve as a non-invasive adjunct tool for preoperative GIST risk stratification without requiring contrast-enhanced CT imaging.