AI-Assisted Colonoscopy: New Research and Guidelines for Clinical Use

Colorectal cancer is now the third most common cancer in the United States and a leading cause of cancer death in both men and women.

Doctors recommend colonoscopies for all adults aged 45 or older to screen for colorectal cancer. During the procedure, gastroenterologists look for and remove polyps that could potentially become malignant. This procedure helps reduce the incidence and mortality rates of colorectal cancer.

“A big part of what we do during a colonoscopy is recognize patterns and then identify pathology,” said Gyanprakash Ketwaroo, MD, MS, associate professor of medicine (digestive diseases) and chief of endoscopy at the West Haven VA. “During a colonoscopy, we are able to thoroughly examine the colon. Studies show that a more detailed review – or a second set of eyes – makes it more likely that we will identify and remove these polyps.”

In 2021, the U.S. Food and Drug Administration (FDA) authorized marketing for the first device that uses artificial intelligence (AI) to help aid endoscopists as they perform colonoscopies. The use of the technology, referred to as computer-aided detection (CADe) for colonoscopies, has grown significantly in recent years.

Dennis Shung, MD, PhD, assistant professor of medicine (digestive diseases) and director of digital health (digestive diseases), is the first author of a recently published meta-analysis in Annals of Internal Medicine, which looked at 44 studies on the effect of AI-assisted colonoscopy systems on adenoma detection rate and advanced colorectal neoplasia detection rate. The study found that doctors using CADe systems found more adenomas than with standard colonoscopies. However, the metanalysis showed that using CADe technology only slightly increased the detection of advanced colorectal neoplasia.

Shung and his colleagues suggest that future studies should look at how AI-aided colonoscopy impacts the incidence of interval colorectal cancer. They also recommend future studies randomize colonoscopists rather than patients.

“It could be that the benefits of CADe are greater for patients who are at higher risk for colorectal cancer,” said Shung. “We need further research to identify where this technology could have the biggest impact.”

Yale began using the technology at the West Haven VA campus about a year ago. The technology is a small addition to the endoscopy software and hardware. Endoscopists simply press a button for AI assistance. When conducting the colonoscopy, they will see a small green box around areas the technology believes are polyps.

“It’s up to you, the endoscopist, to decide whether you agree,” said Ketwaroo. “We have found it helpful to identify small polyps and to provide a second eye to assist you. Our fellows, in particular, like the technology because it gives them the added confidence to know when something is a polyp and to help remove it.”

Ketwaroo also says the technology has been helpful in the removal of tiny polyps, where it can be challenging to keep the polyp in view. The AI assist can help keep the polyp in view and stabilize it so it is easier to remove.

The American Gastroenterological Association (AGA) recently shared new draft clinical guidelines on using AI-assistance in colonoscopies. The guidelines provide evidence and make a conditional recommendation for using CADe to detect polyps in adults. Shung is the co-first author of the guideline.

The evidence base is still developing, but existing evidence shows there are little to no adverse events from CADe use during colonoscopy and no clinically meaningful increase in time needed to conduct the colonoscopy.

“In developing this guideline, we looked at all available data on both provider and patient feedback. Do providers like using the technology? How do patients feel about the use of AI in their screening?” said Shung. “Input from providers and patients is critical."

The final guideline is expected to be published in March 2025. Shung and his coauthors expect to update the guideline regularly as the technology advances.

“Right now, the technology available is a point solution focused on finding polyps at the point of care, but we know that most endoscopy software companies are looking at platform solutions with a suite of AI-assisted tools to help with screening, detection, and removal,” said Shung. “While FDA authorization for any platform solution is not yet clear, we know we need to remain nimble to provide clinicians with guidance so that they can provide the best possible care for patients.”

Since forming one of the nation’s first sections of hepatology more than 75 years ago and then gastroenterology nearly 70 years ago, Yale School of Medicine’s Section of Digestive Diseases has had an enduring impact on research and clinical care in gastrointestinal and liver disorders. To learn more about their work, visit Internal Medicine: Digestive Diseases.

Soleymanjahi S, Huebner J, Elmansy L, Rajashekar N, Lüdtke N, Paracha R, et al. Artificial Intelligence–assisted colonoscopy for polyp detection. Annals of Internal Medicine. 2024 Oct 22; doi:10.7326/annals-24-00981