Improvement in adenoma detection rate by artificial intelligence-assisted colonoscopy: Multicenter quasi-randomized...

Area of Science:

• Gastroenterology and Endoscopy
• Medical Technology
• Artificial Intelligence in Medicine

Background:

• Adenoma detection rate (ADR) is a critical metric in colonoscopy, yet it exhibits variability among endoscopists.
• Artificial intelligence (AI) has emerged as a potential tool to standardize and improve ADR.
• This study investigates the impact of AI on ADR across different levels of endoscopist experience.

Purpose of the Study:

• To evaluate the effect of AI-assisted colonoscopy on adenoma detection rate (ADR).
• To assess whether AI impacts ADR differently based on endoscopist experience.
• To analyze the performance of AI in the screening colonoscopy subset.

Main Methods:

• A prospective, quasi-randomized, controlled, multicenter trial was conducted.
• Patients undergoing colonoscopy were assigned to either AI-assisted (GI Genius) or conventional colonoscopy.
• Endoscopists were categorized as experts (>1000 procedures) or non-experts (≤1000 procedures).

Main Results:

• AI-assisted colonoscopy significantly increased overall ADR (59.1% vs. 46.6%, P < 0.001).
• The improvement in ADR was significant for experts but not for non-experts.
• AI significantly boosted ADR in screening colonoscopies (74.4% vs. 58.1%, P = 0.003) and increased polyp detection rate (PDR).

Conclusions:

• AI-assisted colonoscopy provides a significant overall increase in ADR.
• The benefits of AI in improving ADR were more pronounced in the screening population.
• The observed increase in polyp detection was attributed to higher adenoma detection, not non-neoplastic resections.