Pancreas Divisum: Depiction with Multi–Detector Row CT

Abstract

To retrospectively evaluate contrast material-enhanced multi-detector row computed tomography (CT) in the depiction of pancreas divisum. This study was approved by the investigational review board. Seventy-seven patients (46 men and 31 women; mean age, 51 years) underwent CT with a four-detector row scanner and endoscopic retrograde pancreatography (ERP). Section thickness was 3.2 mm, and the reconstruction interval was 3 mm. Two radiologists independently evaluated the CT data sets with picture archiving and communication system (PACS) workstations equipped with software for two- and three-dimensional postprocessing reformations; the radiologists were blinded to the clinical and ERP data. Pancreas divisum was diagnosed at CT if what the authors termed the "dominant dorsal duct sign" (the caliber of the dorsal duct was larger than that of the ventral duct) was present and if the dorsal and ventral ducts did not appear to communicate with each other at cine review of images. ERP findings were used as the standard of reference for determining the performance (sensitivity, specificity, positive and negative predictive values) of the radiologists' CT interpretations. Interobserver agreement was measured by using kappa statistics. For four of the 77 patients (5%), both radiologists considered that depiction of the pancreatic duct on CT images was not sufficient to enable evaluation of ductal anatomy. These patients were excluded from further analysis. In the remaining 73 patients, ERP demonstrated pancreas divisum in 10 (14%); both observers made the correct diagnosis in nine of these patients. In addition, one radiologist had one false-positive interpretation, whereas the other radiologist had two false-positive interpretations. Thus, for observer 1, the calculated sensitivity was 90% (95% confidence interval [CI], 60%-98%) and the specificity was 98% (95% CI, 91%-100%). For observer 2, sensitivity was 90% (95% CI, 60%-98%) and specificity was 97% (95% CI, 89%-99%). Interobserver agreement was excellent (kappa = 0.93). CT scans obtained with multi-detector row scanners and interpreted with PACS workstations enable depiction of pancreas divisum. This assessment is possible only when the pancreatic duct is visualized.