Applying artificial intelligence to colorectal screenings significantly reduced missed rates for adenomas.

That’s what researchers at Beth Israel Deaconess Medical Center found in what is the first randomized trial in the U.S. to examine the role of a deep-learning computer-aided detection system in colonoscopy. It is also one of the first randomized trials to study the use of AI intervention in any field of medicine.

While reducing risk of death from colon cancer by more than 60%, detecting adenomas with colonoscopy varies among physicians, with missed rates ranging from 6% to 41%. “Those factors may include how meticulous the physician is in searching for polyps. However, certain precancerous polyps are so subtle and might appear only briefly or peripherally on the endoscopy monitor. We know that these polyps can be missed even during meticulous inspection by the physician. AI has the potential to help with this,” senior author Dr. Tyler Berzin, director of the Advanced Endoscopy Fellowship at BIDMC, told HCB News.

Berzin and his colleagues assigned half of 223 patients at four academic medical centers to undergo standard colonoscopy, immediately followed by computer-aided colonoscopy. The other half received the same procedures in reverse order by the same endoscopist. For those who underwent computer-aided colonoscopy first, the missed rate was just above 20%, compared to 34% for those who received standard first.

Berzin says the next step is to prove that the current research, largely obtained from clinical trials at university and academic centers, will lead to cost-effective clinical benefit in real-world clinical settings. “The other key take-home point is that while traditional colonoscopy already detects precancerous polyps earlier and more effectively than stool DNA tests and "virtual" CT scan colonography, the addition of AI polyp detection tools will likely further extend the advantage of colonoscopy as the gold standard for colorectal cancer screening.”

Colorectal cancer is the third most diagnosed cancer and the second leading cause of cancer-related deaths in the U.S. Those at high-risk have a threefold greater chance of developing colorectal cancer, which is why a three-year follow-up interval of colonoscopy is generally recommended. Many of these high-risk patients, however, may not know much about the benefits of colonoscopy, as evidenced in a 2020 study where 28.6% were unaware of either the need for repeat exams or proper surveillance interval.

The trial was conducted between 2019 and 2020.

The findings were published in Clinical Gastroenterology and Hepatology.

Endoscopy Homepage