Artificial intelligence (AI) can detect smaller colon polyps that may be missed by the human eye during a colonoscopy, according to recent research. In studies evaluating how computer-aided detection (CAD), a form of AI, can assist in colon cancer screening, CAD increased detection rates for small adenomas (polyps) compared to a standard colonoscopy, but it did not surpass the skilled eyes of experienced gastroenterologists in finding larger adenomas.
For a randomized controlled trial published in Annals of Internal Medicine, researchers assigned 3,213 participants to undergo standard colonoscopies or colonoscopies with CAD. In both arms, the screening detected advanced adenomas or colorectal cancer in 34% of patients. Although detection for colonoscopy with CAD was comparable to standard colonoscopy, more suspicious lesions were detected per screening using CAD, particularly polyps smaller than 5 mm. In a systematic review and meta-analysis that evaluated 21 randomized trials representing 18,232 colonoscopies, researchers found that CAD colonoscopies had a higher adenoma detection rate (44%) than standard colonoscopies (36%), again particularly with small polyps.
The researchers cautioned that AI’s increased sensitivity may increase rates of unnecessary removal of nonneoplastic polyps and that further studies should investigate CAD’s impact on colorectal cancer incidence and cancer-related mortality rates.
Use of CAD in colon cancer screening has the potential to enhance workflow and expand access. Shortages of trained specialists are limiting screening availability, and many people who should be undergoing colonoscopies for screening are not. CAD or other strategies, such as specially trained nurse practitioners to perform colonoscopies, may be able to optimize patients’ ability to adhere to colorectal cancer screening recommendations.