Post written by Jonathan Makar, BSc, from The University of Melbourne, Melbourne, Victoria, Australia.

Our study focuses on the impact of computer-aided detection (CADe) systems and their role in improving adenoma detection during colonoscopy.

These novel artificial intelligence (AI) systems aim to address endoscopist recognition failure and improve colonoscopy performance, as they are not subject to human recognition biases, distraction, or fatigue, which each contribute to endoscopist recognition failure.

We felt that it was important to perform this updated systematic review and meta-analysis for a number of reasons. Good-quality randomized controlled trials have been published recently, and we sought to include this newly available data in a larger analysis. Other systematic reviews on the topic have been limited by smaller data sets, the inclusion of studies that used computer-aided quality systems alongside CADe systems, and the analysis of live colonoscopy studies alongside video review. All of these may detract from an accurate perspective on the value these CADe systems provide to endoscopists.

Our study offered the largest subgroup analysis of data from experienced endoscopists, as well as a comparison between the most commonly used CADe systems for adenoma detection, and a sensitivity analysis for healthcare setting (tertiary hospital vs day center).

Our analyses indicated that CADe implementation provided a significant improvement to adenoma detection and miss rates, with some nonsignificant benefit for sessile serrated lesions, but a minor prolongation of withdrawal times resulted. Experienced endoscopists also significantly benefited from CADe use, and improvements in adenoma detection rates were seen across all CADe devices and patient care settings.

In light of these results, it is clear that we have some key next steps: first, optimizing AI systems for sessile serrated lesion detection; second, exploring the impact these AI systems may have on endoscopist training and the potential for endoscopist deskilling; and finally, understanding the need for more real-world data for cost-benefit analyses, but more importantly, do these systems actually provide a tangible reduction in colorectal cancer—associated morbidity and mortality? We hope so, but I guess we will wait and see.

Graphical abstract

Read the full article online.

The information presented in Endoscopedia reflects the opinions of the authors and does not represent the position of the American Society for Gastrointestinal Endoscopy (ASGE). ASGE expressly disclaims any warranties or guarantees, expressed or implied, and is not liable for damages of any kind in connection with the material, information, or procedures set forth.