Evaluation of a real-time computer-aided polyp detection system during screening colonoscopy: AI-DETECT study

Abstract

Background Polyp detection and resection during colonoscopy significantly reduce long-term colorectal cancer risk. Computer-aided detection (CADe) may increase polyp identification but has undergone limited clinical evaluation. Our aim was to assess the effectiveness of CADe at colonoscopy within a bowel cancer screening program (BCSP). Methods This prospective, randomized controlled trial involved all eight screening-accredited colonoscopists at an English National Health Service (NHS) BCSP center (February 2020 to December 2021). Patients were randomized to CADe or standard colonoscopy. Patients meeting NHS criteria for bowel cancer screening were included. The primary outcome of interest was polyp detection rate (PDR). Results 658 patients were invited and 44 were excluded. A total of 614 patients were randomized to CADe (n = 308) or standard colonoscopy (n = 306); 35 cases were excluded from the per-protocol analysis due to poor bowel preparation (n = 10), an incomplete procedure (n = 24), or a data issue (n = 1). Endocuff Vision was frequently used and evenly distributed (71.7 % CADe and 69.2 % standard). On intention-to-treat (ITT) analysis, there was a borderline significant difference in PDR (85.7 % vs. 79.7 %; P = 0.05) but no significant difference in adenoma detection rate (ADR; 71.4 % vs. 65.0 %; P = 0.09) for CADe vs. standard groups, respectively. On per-protocol analysis, no significant difference was observed in these rates. There was no significant difference in procedure times. Conclusions In high-performing colonoscopists in a BCSP who routinely used Endocuff Vision, CADe improved PDR but not ADR. CADe appeared to have limited benefit in a BCSP setting where procedures are performed by experienced colonoscopists.

Artificial Intelligence Applied to Colonoscopy: Is It Time to Take a Step Forward?

“…Some studies have suggested that expert endoscopists may benefit less from CADe [ 35 , 36 ]. However, Repici et al [ 37 ] in a post hoc analysis pooling the data of two RCTs compared a CADe and a control group.…”

Navigating the “Trough of Disillusionment” for CADe Polyp Detection: What Can We Learn About Negative AI Trials and the Physician-AI Hybrid?

“…As the authors note, the dual monitor was shown to identify more adenomas $10 mm and flat polyps, but dual-monitor preference was not standardized before randomization and was not controlled for in the multivariate analysis in a way that accounted for individual provider preference. In fact, we do not have a breakdown of whether provider monitor preference was evenly distributed between the groups, when compared, for instance, with the negative study by Ahmad et al (11) in which providers similarly were able to choose whether to concomitantly use a distal cuff attachment, and, importantly, data were provided to demonstrate even distribution of cuff use between groups. It is already known that CADe may affect gaze patterns (13), so allowing a potentially asymmetric provider preference (1 vs 2 screens) into randomization may have affected results.…”

“…Levy et al (10), in a large retrospective observational study conducted at a single high-volume gastroenterology department in Israel, reported a lower ADR in the CADe group compared with that in a pre-CADe retrospective control (30.3% vs 35.2%, P = 0.001) including a lower polyp detection rate and lower APC. Finally, Ahmad et al (11) published an RCT involving 614 patients randomized to receive CADe-assisted colonoscopy vs standard colonoscopy by 8 experts at a UK Bowel Cancer Screening Program and found no difference in ADR but a higher polyp detection rate. Across these 3 trials, explanations for the lack of observed CADe benefit ranged from the possibility of a “ceiling” effect for polyp detection among high-performing endoscopists to “unconscious degradation in the quality of mucosal exposure, possibly due to a false sense of comfort that CADe would ensure a high-quality examination.” Another sentiment that drives these trials is the very real possibility that RCT results may be overoptimistic because endoscopists are generally unblinded and may be conscious that their performance can influence results (9).…”

Artificial intelligence and the push for small adenomas: all we need?

“…In this issue of Endoscopy, Ahmad et al report on their use of CADe within a bowel cancer screening program [38]. The study is distinct from most other randomized trials in several respects.…”