Development and Validation of a CT Radiomics and Deep Learning-Based Model for Predicting Surgical Difficulty in Laparoscopic Pancreaticoduodenectomy

Background Preoperative assessment of surgical difficulty in laparoscopic pancreaticoduodenectomy (LPD) is essential for both patient selection and surgical planning. This study aims to develop and validate an integrated CT radiomics and deep learning model for preoperative prediction of the surgical difficulty score in LPD. Materials and methods A retrospective cohort of 150 patients who underwent LPD between June 2019 and June 2023 was enrolled. Participants were randomly allocated to a training set (n=105) or a testing set (n=45) in a 7:3 ratio. Hand-crafted radiomics (HCR) features and deep learning-derived radiomics (DLR) features were extracted from portal venous phase CT images, focusing on gross tumor volume (GTV) and gross peri-tumor volume (GPTV). A hybrid prediction model was developed using a support vector machine (SVM) algorithm, with performance evaluated through receiver operating characteristic (ROC) analysis, calibration curves, and decision curve analysis (DCA). Results The combined model demonstrated significantly superior discriminative ability, achieving an area under the curve (AUC) of 0.942 (95% CI: 0.893–0.992) in the training set and 0.848 (95% CI: 0.738–0.958) in the testing set. This performance exceeded both the standalone hand-crafted radiomics model (testing AUC = 0.754) and the deep learning-derived radiomics model (testing AUC = 0.816). DCA further confirmed the clinical utility of the combined model, showing the highest net benefit across threshold probabilities exceeding 20%. Conclusion The novel integrated model combining hand-crafted and deep learning-derived radiomics features enables effective prediction of surgical difficulty in laparoscopic pancreaticoduodenectomy, providing a robust tool for preoperative clinical decision-making.