AI-assisted colonoscopy bests conventional colonoscopy for adenoma detection

The authors report no relevant financial disclosures.

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

Artificial intelligence-assisted colonoscopy improved overall adenoma detection rate compared with conventional colonoscopy, according to research published in Clinical Gastroenterology and Hepatology.

“Because a high adenoma detection rate (ADR) is associated with a lower risk of post-colonoscopy colorectal cancer, significant efforts have been devoted to improve ADR. …In recent years, AI-assisted colonoscopy has gained significant clinical interest since it is expected to reduce adenoma miss rate by overcoming human error in polyp detection,” Hong Xu, MD, PhD, professor of medicine and director of the department of gastroenterology and endoscopy center at The First Hospital of Jilin University in China, and colleagues wrote. “While AI-assisted colonoscopy was reported to have a higher ADR and adenoma per colonoscopy in randomized controlled trials and a lower adenoma miss rate in tandem studies when compared to conventional colonoscopy, none of these studies targeted a purely screening population.”

Adenoma detection rate among patients who underwent:  Group A – Conventional colonoscopy; 32.5% Group B – AI-assisted colonoscopy; 39.9%

In a prospective, parallel randomized controlled trial, Xu and colleagues compared AI-assisted colonoscopy with conventional colonoscopy among 3,059 asymptomatic participants, aged 45 to 75 years, who underwent CRC screening by colonoscopy or fecal immunochemical test. Researchers assigned 1,519 participants to the AI-assisted colonoscopy group and 1,540 to the conventional colonoscopy group. Studied outcomes included overall ADR, mean number of adenomas per colonoscopy (APC) and colonoscopy withdrawal time. Researchers also evaluated ADR according to endoscopist experience (non-expert attending vs. expert attending).

Compared with participants who underwent conventional colonoscopy, participants in the AI-assisted group had a higher overall ADR (39.9% vs. 32.4%), advanced ADR (6.6% vs. 4.9%) and APC (0.59 ± 0.97 vs. 0.45 ± 0.81), although median withdrawal time was longer among those in the AI group (8.3 minutes vs. 7.8 minutes). In a subgroup analysis, AI assistance also improved ADR among expert attending endoscopists (42.3% vs. 32.8%) and non-expert attending endoscopists (37.5% vs. 32.1%).

Researchers further reported that AI-assisted colonoscopy increased the detection of adenomas smaller than 5 mm (16.5% vs. 11.5%) and larger than or equal to 10 mm (6.5% vs. 4.7%), as well as non-pedunculated adenomas (27.6% vs. 21.8%) and adenomas in both the proximal (28.4% vs. 23.8%) and distal (10.6% vs. 7.7%) colon.

However, CRC and sessile serrated lesion detection rates did not differ significantly between groups (0.9% vs. 0.8% and 1.1% vs. 1.3%, respectively).

“This large-scale, multicenter randomized study showed the benefits of AI-assisted colonoscopy in asymptomatic subjects undergoing CRC screening. AI-assisted image analysis has already been applied in mammography for the screening of breast cancer, as well as in 3D low-dose CT for the screening of lung cancer,” Xu and colleagues concluded. “It is time for us to consider generalizing the use of AI-assisted endoscopy in the gastrointestinal tract.”

Source link

Back to top button