Colon Lumen Center Detection Enables Autonomous Navigation of an Electromagnetically Actuated Soft-Tethered Colonoscope

Robotic colonoscopes have been extensively developed for providing alternative solutions to conduct colon screening. Nevertheless, most robotic colonoscopes are still facing the challenge of unintuitive and difficult manipulation, particularly in the process of navigation, which may lead to long procedure duration and high workloads for surgeons. To tackle this problem, this article presents autonomous navigation of an electromagnetically actuated soft-tethered (EAST) colonoscope. By utilizing a deep-learning contour segmentation technique, a novel lumen center detection method combines both the contour and darkest region information of colon images to robustly and accurately estimate the center of the upcoming colon lumen, which provides heading directions for EAST colonoscope to navigate through the colon, while a navigation controller is developed to adjust the heading and locomotion velocity of EAST colonoscope accordingly. Comprehensive experiments are conducted to validate the effectiveness and efficacy of the developed lumen center detection model, with autonomous navigation of EAST colonoscope successfully demonstrated in both a commercialized colon phantom and an ex vivo porcine colon. A user study is conducted to verify the effectiveness of our proposed autonomous endoscopic navigation solution in shortening procedure duration and reducing surgeon workloads. The proposed methods are open source and can also be applied to other robotic colonoscopes, which can promote the clinical translation of robotic colonoscopy.

I. Introduction

As one of the most incident cancers with a high mortality rate, colorectal cancer is also one of the most preventable cancers in the world [1]. The survival rate of colorectal cancer remains more than 90% at an early stage, but declines drastically to less than 12% in case of late diagnosis [2]. Therefore, mass and regular colon screening program is of significance in reducing the mortality rate of colorectal cancer [3]. Robotic colonoscopes have been extensively investigated in the past decade for providing alternative solutions to conduct colon screening in a patient-friendly and surgeon-friendly manner [4]. Different active locomotion mechanisms, including legged propulsion [5], wheeled advancement [6], inchworm-liked movement [7], and magnetic actuation [8], [9], [10], [11], enable their robotic colonoscopes to navigate through the colon using a “front-wheel driven” strategy, which can effectively avoid the forcible insertion in the conventional colonoscopy and, thus, reduce the patient discomfort during the procedure.