Water Exchange Colonoscopy With Artificial Intelligence-assisted Detection

Overview

The goal of this clinical trial is to compare the detected adenoma per colonoscopy (APC) in participants undergoing screening, surveillance, and positive fecal immunochemical test (FIT) or guaiac fecal occult blood test (gFOBT). There will be two arms in this study: WE water control and water plus artificial intelligence (AI). The main question it aims to answer is whether the addition of AI into water exchange (WE) colonoscopy increases APC than WE alone. The control method will use water instead of air inserted into the colon. The study method will use a commercially available AI system plus water during the procedure. Researchers will compare APCs to see if the addition of AI increases detection of adenomas during WE colonoscopy.

This will be a multi-site, multi-national, unblinded investigators, prospective Random Control Trial (RCT). Randomization (1:1 to WE alone and WE-AI) will be based on computer generated random numbers placed inside opaque sealed envelopes. The envelope (in pre-arranged order) will be opened to reveal the code when the colonoscopist is ready to insert the endoscope to begin the examination. This will be a comparison of two different methods with two arms (WE, WE-AI) to see which one is better at detecting adenomas. Patient recruitment will be conducted at three hospitals in Italy and Taiwan: Digestive Endoscopy Unit, CTO Hospital, Iglesias, Italy; Digestive Endoscopy and Gastroenterology Unit, Manzoni Hospital, Lecco, Italy; Evergreen General Hospital, Taoyuan, Taiwan. The Sepulveda Ambulatory Care Center, VAGLAHS, UCLA in the USA will be a non-recruiting participating site. The study period is expected to be 3 years (from November 2023 to October 2026). Patients aged 45-75 y/o at average risk of colorectal cancer who are willing to participate will sign an informed consent before starting the colonoscopy procedure. Separate parallel randomization will be set up at each site, stratified by investigator and type of colonoscopy (primary screening, surveillance, or positive FIT or gFOBT). Mode of sedation will include unsedated, on demand sedation, conscious sedation or full sedation with propofol. Randomization will be carried out by computer-generated sequences using a block design (four participants per block). Control Method: One arm of the study will include sedated/unsedated colonoscopy with water (WE) as the control method. Residual air in the colon will be removed and water will be infused to guide insertion through an airless lumen. Infused water will be removed by suction, along with residual fecal debris, predominantly during insertion. Study method: The other arm entail the addition of a commercially available AI system (CAD-EYE, Fujifilm; Endo-AID, Olympus) to the high-definition colonoscopy system. This arm includes sedated/unsedated colonoscopy with AI detection device plus water (WE).