Computer-aided detection in routine colonoscopy improves ADR by 4.1%
Click Here to Manage Email Alerts
Key takeaways:
- Computer-aided detection resulted in an adenoma detection rate of 37.5% vs. 33.7% in standard colonoscopy.
- No safety issues were reported with computer-aided detection in routine practice.
Computer-aided detection in colonoscopy had a “positive effect” on adenoma detection rate, mean number of adenomas per colonoscopy and polyp detection rate and should be considered when performing routine colonoscopy, researchers reported.
“Artificial intelligence systems have been developed to outperform human vision in polyp detection. Some systems have been evaluated in randomized controlled trials, showing promising results,” David Karsenti, MD, of the Digestive Endoscopy Unit at Clinique Paris-Bercy, and colleagues wrote in The Lancet Gastroenterology & Hepatology. “To our knowledge, sparse prospective data are available on the efficacy of computer-aided detection (CADe) systems in routine colonoscopy, specifically in a nonacademic center.”
In a single-center, randomized controlled trial, Karsenti and colleagues evaluated data from 2,015 participants (mean age, 58.4 years; 51.4% women) who underwent CADe-assisted colonoscopy (n = 1,003) or standard colonoscopy (n = 1,012) between May 2021 and May 2022. Results showed an ADR of 37.5% and 33.7%, respectively, which correlated with an estimated mean absolute difference of 4.1% (95% CI, 0-8.1).
The mean number of adenomas detected per colonoscopy (APC) was 0.89 in the CADe group vs. 0.71 in the standard group (estimated mean absolute difference = 0.17; 95% CI, 0.1-0.25), and the proportion of colonoscopies with at least one polyp (PDR) was 45% vs. 41% (estimated mean absolute difference = 4.3%; 95% CI, 0-8.5). There was “no significant difference” between groups for advanced ADR, proximal serrated PDR or distribution of polyp size.
Researchers reported one bleeding event in the CADe group following the resection of a polyp larger than 2 cm, which was resolved after placement of a hemostasis clip during follow-up colonoscopy 4 hours later.
“Our large prospective randomized controlled trial suggests that CADe has a positive effect on ADR, APC and PDR in routine colonoscopy, even in a nonacademic center, but does not lead to the detection of advanced lesions or lesions that are more difficult to see,” Karsenti and colleagues concluded. “The improvement in ADR seemed to be inversely proportional to the basal ADR of the detectors, with the effect of CADe on ADR being low especially among high detectors. We also confirmed that there are no safety issues with the use of CADe in routine practice.”
They added, “Our results provide justification for the systematic use of CADe in routine colonoscopy.”