Related Experiment Video
Updated: Jan 16, 2026

Author Spotlight: Advancing Early Detection and Treatment of Gastrointestinal Tumors
Published on: February 16, 2024
Noninvasive Diagnostic Method for Gastric Subepithelial Tumors Based on Circularity: A Multicenter Prospective Study
Eriko Koizumi1, Osamu Goto1,2, Teppei Akimoto3,4
1Department of Gastroenterology, Nippon Medical School Graduate School of Medicine, Tokyo, Japan.
Introduction:
Gastric subepithelial tumors (SETs) including gastrointestinal mesenchymal tumors (GIMTs) - which often appear as similar hypoechoic lesions originating from the fourth layer of the gastric wall - are difficult to distinguish with endoscopic ultrasound (EUS). We aimed to prospectively validate the utility of circularity, a surrogate indicator of roundness, as a diagnostic method for SETs.
Methods:
Among 100 patients with potential GIMTs sized 1-5 cm prospectively recruited at two institutions between 2020 and 2023, those who experienced pathological diagnosis were included in the final analysis. One representative EUS image showing the tumors' maximum cross-sectional surface was selected, and the circularity was measured using an image analysis software. The circularity of the leiomyoma and other SETs suspected GIMTs was compared, and the diagnostic performance at the optimal cutoff value was evaluated.
Results:
In the 62 patients, the number of GIST, leiomyomas, and other SETs were 46, 10, and 6, respectively. Circularity was significantly lower in the leiomyoma group than in the other SETs group (0.846 vs. 0.924). The area under the receiver operating characteristic curve of circularity for predicting leiomyomas was 0.822 when the cutoff value was set to 0.869. When the lesion was diagnosed as leiomyoma with a circularity of <0.869, the accuracy, sensitivity, and specificity were 82.3%, 70%, and 84.6%, respectively.
Conclusion:
The noninvasive diagnostic method for SETs based on circularity is useful for excluding leiomyomas in lesions of 1-5 cm in size. This diagnostic method may be a potential adjunctive option for differentiating SETs.

