Search

Disposable gastroscopes have recently been developed to eliminate the risk of infection transmission from contaminated reusable gastroscopes. We compared the performance of disposable and reusable gastroscopes in patients undergoing gastroscopy. Patients requiring gastroscopy were randomized to either the disposable or reusable digital gastroscope group. The primary endpoint was the success rate of photographing customary anatomic sites, with a noninferiority margin of –8%. Secondary endpoints were technical performance factors such as gastroscope imaging quality, maneuverability, gastroscopy completion rate, device failure/defect rate, operating time, and safety. Data were analyzed using the Newcombe-Wilson score method and Fisher exact 2-tailed t test. Of 110 patients, 55 were treated using disposable gastroscopes and 55 using reusable gastroscopes. The success rate for capturing images of customary anatomic sites was 100% in both groups. The average imaging quality score was significantly lower (37.02 ± 3.09 vs 39.47 ± 1.92, P <.001) and the operating time significantly longer (P <.001) in the disposable gastroscope group. No significant differences in maneuverability, gastroscopy completion rate, device failure/defect rate, operating time, or safety were found between the 2 groups. Given the overall safety profile and similar technical performance, disposable gastroscopes represent an alternative to reusable gastroscopes for routine examination, bedside first aid, and some certain circumstances. [ABSTRACT FROM AUTHOR]

• The anus detection method covering healthy and diseased anus. • The anus healthy condition classifier, i.e., judging anus is in healthy or diseased condition. • The image processing pipeline for anal center detection based on a set of known image segmentation methods. • In addition, an autonomous lower digestive tract robot can enter the anus through the target point provided by the tube. Objective : The very first step for an autonomous lower gastrointestinal (GI) tract robot to carry out a diagnostic or a therapeutic task is to enter the lower GI. The natural entry point of the lower GI tract is the anus. Thus, to find the anus center is the very first step before entering the lower GI tract. However, to the authors' knowledge, there doesn't exist any results of detection of anus center. Methods : An image processing pipeline is proposed by combing several classical image methods including Otsu's method, the multiscale method and the threshold method and two deep neural networks, including Mask R-CNN and Inception-V3. Also, as a complementary result, the classification of healthy and diseased anus is determined by another Inception-V3, for healthy anus. Results : The positional error of the center detection by the proposed workflow is 2.15% averagely compared to the diagonal length, which is at the same level to that ten experienced proctologists. The proposed anal center detection method is applicable for both hairy anus and hairless anus. Also, the approach is valid for the anus in several common perianal diseases including perianal eczema, the mixed hemorrhoids, the anal fistula, the thrombotic external hemorrhoids, the internal hemorrhoids, and the external hemorrhoid. Conclusion : Anus center is detected by the proposed method with a similar accuracy to human doctors. Significance : This study provides the first solution for the anus center detection, enabling autonomous lower gastrointestinal tracts robot to enter anus without human guidance. [ABSTRACT FROM AUTHOR]