Computer Vision News - June 2022

41 CRAS 2022 the urinary tract with a camera at the tooltip. It can be controlled autonomously or manually with a joystick. It has an autonomous visual-motor system that tracks the center of the lumen and moves the device’s tip towards it, which helps avoid the risk of lesions. The clinician always has the option to revert to manual control. “ We use FBG sensors together with EM trackers to localize the whole length of the device and provide this information back to the user, ” Martina explains. Ureteroscopy is not an easy procedure to perform. Navigation inside the urinary tract is complex, and it depends on the user’s experience as to whether it will be successful. The operator must avoid touching the walls of the lumen inside the ureter as this could cause lesions. Localization is achieved by performing repetitive X-rays, which can be harmful to both patient and user. Martina and Sanat’s award-winning work includes a steerable robotic device – developed by Chun-Feng Lai , one of the project’s co-authors – which travels along