Duodenal papilla radiomics-based prediction model for post-ERCP pancreatitis using machine learning: a retrospective multicohort study

Abstract

Background and Aims

The duodenal papillae are the primary and essential pathway for ERCP, greatly determining its complexity and outcome. We investigated the association between papilla morphology and post-ERCP pancreatitis (PEP) and constructed a robust model for PEP prediction.

Methods

We retrospectively enrolled patients who underwent ERCP in 2 centers from January 2019 to June 2022. Radiomic features of the papilla were extracted from endoscopic images with deep learning. Potential predictors and their importance were evaluated with 3 machine learning algorithms. A predictive model was developed using best subset selection by logistic regression, and its performance was evaluated in terms of discrimination, calibration, and clinical utility based on the area under curve (AUC) of the receiver-operating characteristic curve, calibration curve, and clinical decision curve, respectively.

Results

From 2 centers, 2038 and 334 ERCP patients were enrolled in this study with PEP rates of 7.9% and 9.6%, respectively. The radiomic score was significantly associated with PEP and showed great diagnostic value (AUC, .755-.821). Six hub predictors were selected to conduct a predictive model. The radiomics-based model demonstrated excellent discrimination (AUC, .825-.857) and therapeutic benefits in the training, testing, and validation cohorts. The addition of the radiomic score significantly improved the diagnostic accuracy of the predictive model (net reclassification improvement, .151-.583 [P < .05]; integrated discrimination improvement, .097-.235 [P < .001]).

Conclusions

The radiomic signature of the papilla is a crucial independent predictor of PEP. The papilla radiomics-based model performs well for the clinical prediction of PEP.