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

Area of Science:

• Medical Imaging
• Radiomics
• Metabolic Syndrome Research

Background:

• Early identification of severe acute pancreatitis (SAP) in metabolic syndrome (MetS) patients is vital for prognosis and timely intervention.
• Conventional scoring systems (BISAP, MRSI) have limited accuracy for early SAP prediction.
• MRI-based radiomics can quantify subtle image features for noninvasive disease assessment, potentially improving SAP prediction when combined with clinical data.

Purpose of the Study:

• To develop and validate a predictive model for SAP in MetS patients.
• The model integrates MRI T2WI radiomics features with clinical indicators.
• To enhance the early prediction accuracy of SAP in this patient cohort.

Main Methods:

• Retrospective study of 188 acute pancreatitis (AP) patients with MetS.
• Radiomics features extracted from MRI T2WI using PyRadiomics after 3D Slicer delineation.
• A random forest classifier built a radiomics model; logistic regression identified clinical predictors for a combined model.

Main Results:

• Seven radiomics features were selected after dimensionality reduction.
• Length of hospital stay and serum calcium identified as independent clinical predictors.
• The combined model achieved high AUCs (0.97 training, 0.979 test), outperforming individual clinical, radiomics, BISAP, and MRSI models.

Conclusions:

• A combined model integrating MRI T2WI radiomics and clinical features offers robust prediction of SAP in MetS patients.
• This approach demonstrates significant clinical value for early detection and intervention.
• The findings support the model's potential utility in clinical practice for improving SAP management.