Artificial Intelligence for Adenoma Detection Rate: Simply the Best?
Douglas K. Rex, MD, MASGE, reviewing Aziz M, et al. J Clin Gastroenterol 2022 Nov 28.
This study assessed 94 randomized controlled trials with 61,172 patients to determine how artificial intelligence (AI) impacts the adenoma detection rate (ADR). In short, AI was better than everything else for increasing ADR, including dye-based chromoendoscopy (relative risk [RR], 1.22), endocap (RR, 1.32), Endocuff (RR, 1.19), Endocuff Vision (RR, 1.26; Olympus America, Center Valley, Penn), narrow-band imaging (RR, 1.33), linked color imaging (RR, 1.21), and water exchange (RR, 1.22), as well as 5 other more seldom used modalities.
![d-rex](/wp-content/uploads/d-rex.jpg)
COMMENTThis type of analysis does not replace head-to-head comparisons. Therefore, is AI simply the best detection adjunct? The fairly consistent results of 10% to 12% increases in ADR across parallel arm studies and a 50% reduction in adenoma miss rates in tandem studies suggest that AI is the most effective lesion-highlighting method. It’s likely also the easiest to use, though not the least expensive.
The more important question is whether AI can be enhanced by combining it with other modalities. The logical device to use along with AI is a mucosal exposure device. Available evidence suggests that AI combined with Endocuff Vision has additive benefits for increasing ADR, as do AI computer-aided detection programs plus AI computer-aided quality assessment programs that assess and correct endoscopist examination techniques.
Note to readers: At the time we reviewed this paper, its publisher noted that it was not in final form and that subsequent changes might be made.
CITATION(S)
Aziz M, Haghbin H, Sayeh W, et al. Comparison of artificial intelligence with other interventions to improve adenoma detection rate for colonoscopy: a network meta-analysis. J Clin Gastroenterol 2022 Nov 28. (Epub ahead of print) (https://doi.org/10.1097/mcg.0000000000001813)