Value of Dual-Energy CT Perfusion Analysis in Patients with Acute Pancreatitis: Correlation and Discriminative Diagnostic Accuracy with Varying Disease Severity

Abstract

Background: This study investigates the correlation and discriminative diagnostic accuracy of dual-energy CT (DECT)-derived imaging biomarkers in patients with acute pancreatitis of varying severity. Methods: In this retrospective study, we included 51 patients with acute pancreatitis who had undergone portal-venous phase DECT of the abdomen. Three blinded readers independently performed region-of-interest measurements on DECT images in the inflammatory pancreatic parenchyma. The correlation between modified CT severity index (CTSI) and quantitative imaging parameters was investigated using Pearson correlation coefficient. We performed receiver operator curve (ROC) analysis to assess diagnostic accuracy of the quantitative image parameters for the differentiation between mild/moderate versus severe acute pancreatitis. The optimal discriminative cut-off value to diagnose severe acute pancreatitis was determined using the Youden index. Results: Moderate correlations were found between CTSI scores and iodine density (Pearson’s correlation coefficient r = −0.65; p < 0.001), as well as attenuation (r = −0.55; p < 0.001) and normalized iodine uptake (r = −0.50; p < 0.001). ROC curve analysis revealed highest ability to differentiate mild/moderate from severe acute pancreatitis for iodine density (AUC = 0.86, 95% confidence interval 0.75 to 0.97). An optimal iodine density threshold of ≤1.63 mg/mL was found to indicate severe acute pancreatitis with a sensitivity of 81.3% and specificity of 77.1%. Conclusion: DECT-derived iodine density correlates with acute pancreatitis severity and may facilitate prediction of severe acute pancreatitis.