Subtraction color map of contrast-enhanced and unenhanced CT for the prediction of pancreatic necrosis in early stage of acute pancreatitis

The objective of our study was to evaluate the accuracy of subtraction color-map images created from contrast-enhanced CT (CECT) and unenhanced CT for the diagnosis of pancreatic necrosis in the early stage of acute pancreatitis.Forty-eight patients underwent unenhanced CT and CECT within 72 hours from the onset of acute pancreatitis. Subtraction color-map images were created from unenhanced CT and CECT using a 3D nonrigid registration method. Three radiologists reviewed two image sets: CECT alone and subtraction color-map images in conjunction with CECT. Readers evaluated each image set for the presence of pancreatic necrosis. The reference standard for pancreatic necrosis was CT or MRI 1 week or more after the onset of acute pancreatitis. The performance of each image set for the prediction of pancreatic necrosis was calculated and compared using the McNemar test.Eleven of the 48 patients developed pancreatic necrosis. There were no technical failures creating the subtraction images. The sensitivity, specificity, and accuracy for predicting pancreatic necrosis with CECT were 64%, 97%, and 90%, respectively, for reader 1; 73%, 87%, and 83% for reader 2; and 73%, 87%, and 83% for reader 3. The sensitivity, specificity, and accuracy for predicting pancreatic necrosis with the subtraction color maps were 100%, 100%, and 100%, respectively, for reader 1; 100%, 95%, and 96% for reader 2; and 82%, 92%, and 90% for reader 3. Accuracy significantly improved with the addition of subtraction color maps compared with CECT alone for reader 1 (p = 0.03) and reader 2 (p = 0.02) but not for reader 3 (p = 0.37).A subtraction color map is accurate in the diagnosis of pancreatic necrosis in the early stage of acute pancreatitis.