Purpose: This study aims to assess the diagnostic accuracy of virtual gastroscopy obtained by 320-row computed tomography (CT) examination in differentiating benign from malignant gastric ulcers (GUs).
Materials and methods: 49 patients (30 M, 19 F, mean age 58.6) with endoscopic and histological diagnosis of GU underwent CT examination. A hypotonizing drug was administered and gastric walls were distended by air in order to perform virtual endoscopy. Based on morphological features, GUs were subdivided into benign or malignant forms by two blinded radiologists. Interobserver agreement was evaluated using Cohen’s kappa (κ) test. CT results were then compared with endoscopic and histological findings, having the latter as the reference standard.
Results: Thirty-five out of 49 patients (71%) were affected by malignant ulcers, while in the remaining 14 cases diagnosis of benign GU was made. Virtual gastroscopy showed diagnostic accuracy, sensitivity, and specificity values of 94%, 91%, and 100%, respectively, in differentiating benign from malignant ulcers. Almost perfect agreement between the two readers was found (κ = 0.86).
Conclusion: CT virtual gastroscopy improves the identification of GUs and allows differentiating benign from malignant forms.