The Effect of Computer-Aided Device on Adenoma Detection Rate in Different Implement Scenarios: A Real-World Study

Area of Science:

• Gastroenterology
• Artificial Intelligence in Medicine
• Medical Imaging Analysis

Background:

• Real-world efficacy of computer-aided polyp detection (CADe) for adenoma detection rate (ADR) has shown variability.
• Unmeasured factors in AI-human interaction, such as monitor display methods, require investigation.
• This study validates CADe's real-world effectiveness and assesses monitor approach impact.

Purpose of the Study:

• To validate the real-world effectiveness of computer-aided polyp detection (CADe) in improving adenoma detection rates (ADR).
• To assess the impact of different monitor approaches (dual vs. single) on CADe's performance.
• To evaluate the overall utility of CADe in routine colonoscopy settings.

Main Methods:

• Retrospective propensity score-matched cohort study using routine data from a tertiary endoscopy center.
• Comparison of pre-CADe, dual-monitor CADe-assisted, and single-monitor CADe-assisted groups.
• Adenoma detection rate (ADR) was the primary outcome measure.

Main Results:

• CADe-assisted groups demonstrated a trend towards increased ADR compared to the pre-CADe period.
• Dual-monitor and single-monitor CADe assistance showed increased ADR (OR 1.178 and 1.094, respectively).
• No significant difference in ADR was observed between dual and single monitor setups, though dual showed a tendency (OR 1.069).

Conclusions:

• Computer-aided polyp detection (CADe) holds significant potential for enhancing adenoma detection rates during colonoscopy in clinical practice.
• Monitor configuration (dual vs. single) does not appear to significantly alter the effectiveness of CADe.
• Further research into AI-clinician interaction factors is recommended for optimal CADe implementation.