Utilizing MRI Radiomics and Clinical Features to Predict Severe Acute Pancreatitis in Patients with Metabolic Syndrome.

BACKGROUND

Early identification of severe acute pancreatitis (SAP) in patients with metabolic syndrome (MetS) is crucial for improving prognosis and guiding timely intervention. Conventional scoring systems such as the bedside index for severity in acute pancreatitis (BISAP) and magnetic resonance severity index (MRSI) show limited accuracy for early prediction. MRI-based radiomics offers a noninvasive approach to quantify subtle image features that may reflect underlying disease heterogeneity. Integrating radiomics with clinical indicators may enhance prediction of SAP progression in MetS patients.

PURPOSE: To develop and validate a predictive model combining MRI T2WI radiomics and clinical features to predict SAP occurrence in MetS patients.

PATIENTS AND

METHODS

This retrospective study included 188 patients with acute pancreatitis (AP) and MetS, classified into severe (31 patients) and non-severe (157 patients) groups according to the 2012 revised Atlanta consensus. Regions of interest were delineated using 3D Slicer, and radiomics features were extracted via PyRadiomics. Features were normalized and selected using select K-best and least absolute shrinkage and selection operator (LASSO). A random forest classifier constructed the radiomics model, while binary logistic regression identified independent clinical predictors to form a combined model. Model performance and clinical utility were evaluated using the area under the curve (AUC), the DeLong test, and decision curve analysis (DCA).

RESULTS

Seven radiomics features were selected following dimensionality reduction. Binary logistic regression identified length of hospital stay and serum calcium as independent clinical risk factors. The combined model achieved AUCs of 0.97 and 0.979 in training and test sets, respectively, outperforming the clinical, radiomics, BISAP, and MRSI models.

CONCLUSION

The combined model integrating MRI T2WI radiomics with clinical features provides robust and clinically valuable prediction of SAP in MetS patients, supporting its potential value for early clinical intervention.