Three-dimensional MDCT Gastrography Compared With Axial CT for the Detection of Early Gastric Cancer

OBJECTIVE: To assess the diagnostic performance of multidetector-row computed tomography (CT) 3-dimensional CT gastrography (3-D CTG) to detect early gastric cancer (EGC) compared with axial CT images alone. METHODS: Contrast-enhanced multidetector-row CT scanning was performed on 39 patients with histopathologically proven EGC. By using volume-rendering technique, CTG images were created and interval reviews of both the axial images without and with 3-D CTG images were performed independently by 2 radiologists retrospectively. The stomach was divided into 4 segments (ie, the cardia or fundus, body, angle, and antrum). For 156 gastric segments, the radiologists determined the presence of the lesion using a 5-point confidence level. The radiologists' performance for the lesion detection was evaluated by means of receiver operating characteristic analysis. Interobserver agreement was also analyzed. The ability of CTG to reveal the extent and the morphological features of the lesions was also evaluated and compared with the results of conventional studies. RESULTS: Histopathologically, 41 EGCs were identified in 39 patients: 1 lesion was located in the fundus, 16 in the body, 9 in the angle, and 15 in the antrum. When 3-D CTG images were used, the receiver operating characteristic curve (AZ) analysis revealed a significant improvement in the diagnostic performance of both reviewers (AZ using axial CT images only, 0.608 and 0.602 for reviewers 1 and 2, respectively; AZ using CTG images, 0.821 and 0.822 for reviewers 1 and 2, respectively) (P < 0.05). The 3-D CTG also improved the sensitivity from 27% to 73% for reviewer 1 and from 29% to 76% for reviewer 2 (P < 0.05), as compared with the sensitivity when using axial CT images only. In addition, almost perfect agreement was achieved for CTG (weighted kappa, 0.836), whereas there was only moderate agreement for the axial CT images (weighted kappa, 0.445). The CTG provided information similar to that obtained by barium study and endoscopy in 51.6% and 59.5%, respectively, of the cases. CONCLUSIONS: The combined interpretation of axial and 3-D CTG was significantly better for detecting EGC, with a diagnostic confidence higher than that using axial CT imaging alone.