AI-assisted GI Genius ‘had no impact’ on serrated polyp detection in positive stool tests
Click Here to Manage Email Alerts
Key takeaways:
- Implementing computer-aided detection for colonoscopy had “no impact” on the rate of detection for serrated polyps.
- The system only marginally improved adenoma detection rate.
PHILADELPHIA—An AI-assisted endoscopy module minorly improved adenoma detection during colonoscopy, but “had no impact” on serrated polyp detection in patients with positive stool tests, a presenter said at the ACG Annual Scientific Meeting.
Although computer-aided polyp detection (CADe) is known to improve polyp identification during screening and surveillance colonoscopy, “there is little known about how AI impacts polyp detection in higher-risk patients, such as those with positive stool-based testing,” Rajesh N. Keswani, MD, of the division of gastroenterology and hepatology at Northwestern Medicine, told Healio.
To assess the influence of CADe on this patient subset, Keswani and colleagues retrospectively examined how the inclusion of the GI Genius (Medtronic) module affected polyp detection among patients referred following a positive stool-based testing, either a multitarget stool DNA test (MT-sDNA; Cologuard, Exact Sciences) or fecal immunological testing (OC-Auto FIT; Polymedco).
Four CADe units were placed in 12 rooms of an endoscopy unit at a single urban academic center, in which 51 providers would rotate between the rooms daily and perform colonoscopies with or without the CADe system, depending on its availability.
The primary outcome for the study was the effect of CADe on the adenoma detection rate (ADR), serrated polyp detection rate and neoplasia detection rate (NDR). Keswani and colleagues noted that, over 6-months, providers performed 15,719 colonoscopies, with 194 patients referred for having positive MT-sDNA testing and 82 for positive FIT results.
According to study results, use of CADe only marginally increased the rate of neoplasia detection (74.6% vs. 67.9%, P=.4) and adenoma detection (70.1% vs. 60.3%, P=.2), while exhibiting “no impact” on serrated polyp detection (23.9% vs. 24.9).
“In our single center analysis, AI had a non-significant improvement in the detection of adenomas (approximately 10%), but had no impact on serrated polyp detection,” Keswani told Healio. “This research lends further evidence to the importance of AI in ensuring high-quality colonoscopy. We would like to better understand the impact of AI in the detection of advanced neoplasia in this patient population.”
Collapse
Keswani R, et al. Adoption of a computer-aided detection system may improve polyp detection in patients with positive stool-based testing; Oct. 25-30, 2024; Philadelphia (hybrid meeting).
Click Here to Manage Email Alerts