Objectives: To compare the respective capabilities of the arterial, the portal, and the combined set in the detection and localization of acute gastrointestinal (GI) bleeding with 64-section computed tomography (CT).
Methods: A total of 46 patients with acute GI bleeding and who had undergone both 64-section CT and digital subtraction angiography were included in this study. The results of angiography were used as a reference standard. Two radiologists independently reviewed the 3 sets of CT images (arterial set, the unenhanced and arterial-phase images; portal set, the unenhanced and portal venous-phase images; combined set, the unenhanced and arterial-phase and portal venous-phase images). The diagnostic accuracy was assessed by a receiver operating characteristic analysis.
Results: For each observer, the Az values were 0.915 and 0.931 for the arterial set, 0.903 and 0.933 for the portal set, and 0.919 and 0.911 for the combined set, respectively. The differences were not statistically significant among the 3 data sets for each observer (P > 0.05). Both observers correctly detected the bleeding site in 81.3% and 84.4% on the arterial set, in 81.3% and 84.4% on the portal set, and in 84.4% and 84.4% on the combined set, respectively.
Conclusions: Using 64-section CT, the diagnostic performance was not different among the arterial, the portal, and the combined set for the detection and localization of acute GI bleeding.