A CT-Based Decision Tree Model for Differentiating Sub-3 cm Gastric Ectopic Pancreas from Gastrointestinal Stromal Tumors

Objective To develop a CT-based decision tree model integrating clinical and imaging features for the preoperative differentiation of gastric ectopic pancreas (GEPs) and gastrointestinal stromal tumors (GISTs) with a long diameter of less than 3 cm. Methods This retrospective study included 86 patients with pathologically confirmed GEPs (n = 26) or GISTs (n = 60) from two centers between 2014 and 2024. All patients underwent contrast-enhanced CT prior to surgery. The dataset was divided into training (70%) and testing (30%) sets. Significant variables were selected via univariate analysis and logistic regression, and a decision tree model was constructed. Diagnostic performance was assessed using ROC analysis, AUC, optimal cutoff, and calibration curve. SHAP was used for feature importance interpretation. Results The decision tree model identified four key variables: age (clinical factor) and three CT features: ratio of lesion-to-pancreas attenuation in the arterial phase(A2), lesion long-to-short diameter ratio (LD/SD ratio), and intralesional low attenuation (ILA). The model, based on these four features, achieved an AUC of 0.935 (95% CI, 0.869–0.979), with sensitivity of 70.0% and specificity of 84.0%. Calibration analysis showed good agreement between estimated and true values. Conclusions The CT-based decision tree model, integrating four clinical and CT features, provides an effective and interpretable tool for differentiating GEPs from GISTs with a long diameter of less than 3 cm, demonstrating strong diagnostic performance.