Diagnostic performance of virtual gastroscopy using MDCT in early gastric cancer compared with 2D axial CT: Focusing on interobserver variation

OBJECTIVE. The objective of our study was to assess the diagnostic performance of virtual gastroscopy using MDCT for the detection of early gastric cancer (EGC) compared with 2D axial CT, focusing on interobserver reliability. MATERIALS AND METHODS. During an 11-month period, we performed CT examinations of 94 consecutive patients with EGC and a control group composed of 68 patients without EGC. Three radiologists retrospectively interpreted the 2D axial CT and virtual gastroscopy images. Diagnostic performances were compared within each observer using the area under the receiver operating characteristic curve (A(z)). Sensitivity and specificity were also calculated for each individual observer. The simple kappa statistic was used to evaluate interobserver reliability in the detection of EGC. RESULTS. The diagnostic performance for overall lesion detection in patients with EGC was significantly higher with virtual gastroscopy (A(z) = 0.829-0.885) than with 2D axial CT (A(z) = 0.734-0.793) (p < 0.001). The sensitivity and specificity of virtual gastroscopy for EGC were 78.7-84.0% and 83.8-91.2%, respectively. The sensitivity and specificity of 2D axial CT for EGC were 62.8-69.2% and 83.8-92.7%, respectively. Virtual gastroscopy showed a higher sensitivity for EGC than 2D axial CT (p < 0.001). The interobserver reliabilities showed moderate to substantial agreement (kappa = 0.40-0.74 for 2D axial CT, kappa = 0.57-0.71 for virtual gastroscopy). CONCLUSION. Virtual gastroscopy showed excellent results with a good interobserver reliability for the detection of EGC compared with 2D axial CT.