Artificial intelligence for identification and characterization of colonic polyps

Area of Science:

• Gastroenterology and Endoscopy
• Artificial Intelligence in Medicine
• Colorectal Cancer Screening

Background:

• Colonoscopy is the gold standard for colorectal cancer screening, but its effectiveness is operator-dependent, leading to missed polyps and interval cancers.
• Current strategies like 'resect-and-discard' and 'diagnose-and-leave' for diminutive polyps are hampered by suboptimal optical biopsy accuracy in practice.
• Interval colorectal cancers remain a concern due to missed lesions during colonoscopy.

Purpose of the Study:

• To review the recent literature on artificial intelligence (AI) applications for colorectal polyp detection and characterization.
• To evaluate the impact of AI on improving colonoscopy quality metrics, including adenoma detection rates and diagnostic accuracy.
• To discuss the limitations of current AI technologies and outline future directions for clinical implementation.

Main Methods:

• Review of recent scientific literature focusing on AI-assisted computer-aided detection and diagnosis in colonoscopy.
• Analysis of studies evaluating AI's performance in real-time colonoscopy for polyp identification and characterization.
• Examination of AI's role in improving optical biopsy accuracy and reducing diagnostic time.

Main Results:

• AI-assisted colonoscopy has demonstrated increased adenoma detection rates and improved optical biopsy accuracy.
• AI tools have shown potential in reducing endoscopist withdrawal time and expediting polyp diagnosis.
• These advancements represent promising steps towards standardizing colonoscopy quality and supporting new polyp management strategies.

Conclusions:

• AI holds significant potential to enhance the accuracy and efficiency of colonoscopy for colorectal cancer screening.
• AI-assisted systems can improve polyp detection, characterization, and aid in the implementation of 'resect-and-discard' and 'diagnose-and-leave' strategies.
• Further research and regulatory approval are necessary to address real-world application challenges before widespread clinical adoption of AI in colonoscopy.