Current status and limitations of artificial intelligence in colonoscopy

Area of Science:

• Endoscopy
• Artificial Intelligence
• Medical Imaging

Background:

• Deep learning-based artificial intelligence (AI) for polyp detection (CADe) and characterization (CADx) is nearing clinical integration.
• Computer-aided detection (CADe) has demonstrated potential in randomized controlled trials.
• Further advancements are necessary to optimize AI for polyp characterization (CADx).

Purpose of the Study:

• To provide a comprehensive overview of current AI applications in colonoscopy.
• To review the performance of AI systems in screening colonoscopies.
• To discuss the limitations and legal aspects of AI in colonoscopy.

Main Methods:

• A literature search was conducted to identify key studies on AI in colonoscopy.
• The review focuses on prospective trials for CADe and the research status of CADx.

Main Results:

• AI for polyp detection (CADe) has shown potential in initial prospective trials.
• Research in AI for polyp characterization (CADx) is ongoing.
• Current limitations and legal considerations for AI systems in colonoscopy were identified.

Conclusions:

• AI holds significant promise for improving colonoscopy outcomes.
• Continued research and development are crucial for the widespread clinical adoption of AI in colonoscopy.
• Addressing system limitations and legal frameworks is essential for successful implementation.