AI-Driven Colonoscopy Aid Module Boosts Premalignant Polyp Detection

Results from the British COLO-DETECT trial add to the growing body of evidence supporting the use of artificial intelligence (AI)-aided colonoscopy to increase premalignant colorectal polyp detection in routine colonoscopy practice.

Colin J. Rees, a professor of gastroenterology in the Faculty of Medical Sciences at Newcastle University in Newcastle upon Tyne, England, and colleagues compared the real-world clinical effectiveness of computer-aided detection (CADe)-assisted colonoscopy using an “intelligent” module with that of standard colonoscopy in a study in The Lancet Gastroenterology & Hepatology.

They found the GI Genius Intelligent Endoscopy Module (Medtronic Inc.) increased the mean number of adenomas detected per procedure and the adenoma detection rate, especially for small, flat (type 0-IIa) polyps, and sessile serrated lesions, which are more likely to be missed.

“Missed sessile serrated lesions disproportionately increase the risk of post-colonoscopy colorectal cancer, thus the adoption of GI Genius into routine colonoscopy practice could not only increase polyp detection but also reduce the incidence of post-colonoscopy colorectal cancer,” the investigators wrote.

“AI is going to have a major impact upon most aspects of healthcare. Some areas of medical practice are now well established, and some are still in evolution,” Prof Rees, who is also president of the British Society of Gastroenterology, said in an interview. “Within gastroenterology, the role of AI in endoscopic diagnostics is also evolving. The COLO-DETECT trial demonstrates that AI increases detection of lesions, and work is ongoing to see how AI might help with characterization and other elements of endoscopic practice.”

Study Details

The multicenter, open-label, parallel-arm, pragmatic randomized controlled trial was conducted at 12 National Health Service hospitals in England. The study cohort consisted of adults ≥ 18 years undergoing colorectal cancer (CRC) screening or colonoscopy for gastrointestinal symptom surveillance owing to personal or family history.

Recruiting staff, participants, and colonoscopists were unmasked to allocation, whereas histopathologists, cochief investigators, and trial statisticians were masked.

CADe-assisted colonoscopy consisted of standard colonoscopy plus the GI Genius module active for at least the entire inspection phase of colonoscope withdrawal.

The primary outcome was mean adenomas per procedure (total number of adenomas detected divided by total number of procedures). The key secondary outcome was adenoma detection rate (proportion of colonoscopies with at least one adenoma).

From March 2021 to April 2023, the investigators recruited 2032 participants, 55.7% men, with a mean cohort age of 62.4 years and randomly assigned them to CADe-assisted colonoscopy (n = 1015) or to standard colonoscopy (n = 1017). Of these, 60.6% were undergoing screening and 39.4% had symptomatic indications.

Mean adenomas per procedure were 1.56 (SD, 2.82; n = 1001 participants with data) in the CADe-assisted group vs 1.21 (n = 1009) in the standard group, for an adjusted mean difference of 0.36 (95% CI, 0.14-0.57; adjusted incidence rate ratio, 1.30; 95% CI, 1.15-1.47; P

Adenomas were detected in 555 (56.6%) of 980 participants in the CADe-assisted group vs 477 (48.4%) of 986 in the standard group, representing a proportion difference of 8.3% (95% CI, 3.9-12.7; adjusted odds ratio, 1.47; 95% CI, 1.21-1.78; P

As to safety, adverse events were numerically comparable in both the intervention and control groups, with overall events 25 vs 19 and serious events 4 vs 6. On independent review, no adverse events in the CADe-assisted colonoscopy group were related to GI Genius.

Offering a US perspective on the study, Nabil M. Mansour, MD, an associate professor and director of the McNair General GI Clinic at Baylor College of Medicine in Houston, said GI Genius and other CADe systems represent a significant advance over standard colonoscopy for identifying premalignant polyps. “While the data have been mixed, most studies, particularly randomized controlled trials have shown significant improvements with CADe in detection both terms of in adenomas per colonoscopy and reductions in adenoma miss rate,” he said in an interview.

He added that the main utility of CADe is for asymptomatic patients undergoing average-risk screening and surveillance colonoscopy for CRC screening and prevention, as well as for those with positive stool-based screening tests, “though there is no downside to using it in symptomatic patients as well.” Though AI colonoscopy likely still stands at

The main question, Mansour cautioned, is whether increased detection of small polyps will actually reduce CRC incidence or mortality, and it will likely be several years before clear, concrete data can answer that.

“Most studies have shown the improvement in adenoma detection is mainly for diminutive polyps

This study was supported by Medtronic Inc.

Rees reported receiving grant funding from ARC Medical, Norgine, Medtronic, 3-D Matrix, and Olympus Medical, and has been an expert witness for ARC Medical. Other authors disclosed receiving research funding, honoraria, or travel expenses from Medtronic or other private companies.

Mansour had no competing interests to declare.