Artificial Intelligence-Assisted Colonoscopy for Polyp Detection : A Systematic Review and Meta-analysis.

Area of Science:

• Gastroenterology
• Medical Technology
• Endoscopy

Background:

• Randomized clinical trials (RCTs) comparing computer-aided detection (CADe) system-enhanced colonoscopy with conventional colonoscopy indicate improved adenoma detection rates (ADR) and reduced adenoma miss rates (AMR).
• However, the impact of CADe-enhanced colonoscopy on detecting advanced colorectal neoplasia (ACN) remains unclear.

Purpose of the Study:

• To systematically review and compare the performance of CADe-enhanced colonoscopy versus conventional colonoscopy.
• The review aims to clarify the effectiveness of CADe systems in detecting colorectal neoplasia.

Main Methods:

• A systematic search of multiple databases (Cochrane Library, Google Scholar, EMBASE, MEDLINE, PubMed, Scopus, Web of Science) was conducted up to February 2024.
• Included studies were published RCTs comparing CADe-enhanced and conventional colonoscopy.
• Primary outcomes included average adenoma per colonoscopy (APC) and ACN per colonoscopy; secondary outcomes were ADR, AMR, and ACN detection rate (ACN DR).

Main Results:

• Forty-four RCTs involving 36,201 patients were analyzed.
• CADe-enhanced colonoscopies demonstrated higher APC (0.98 vs. 0.78) and ADR (44.7% vs. 36.7%) compared to conventional colonoscopies.
• While average ACN per colonoscopy was similar (0.16 vs. 0.15), the ACN detection rate (ACN DR) was significantly higher with CADe (12.7% vs. 11.5%).
• CADe use led to a slight increase in nonneoplastic polyp resections and total withdrawal time.

Conclusions:

• CADe-enhanced colonoscopy significantly increases adenoma detection rates.
• The technology shows no significant difference in the detection of advanced colorectal neoplasia per colonoscopy, although the detection rate is slightly improved.
• Performance estimates were potentially affected by study heterogeneity and the inability to blind endoscopists.