Impact of artificial intelligence on colorectal polyp detection

Area of Science:

• Gastroenterology
• Medical Imaging
• Artificial Intelligence

Background:

• Colorectal cancer (CRC) incidence and mortality have decreased due to colonoscopy and polypectomy.
• However, up to 25% of colorectal neoplasia are missed, contributing to interval CRC.
• Missed lesions stem from recognition failure and inadequate mucosal exposure.

Purpose of the Study:

• To review recent literature on the performance of Computer Assisted Detection (CADe) for colorectal polyps.
• To evaluate the effectiveness of AI in real-time colonoscopic examinations.

Main Methods:

• Overview of recent literature, including retrospective studies and randomized clinical trials.
• Focus on AI systems designed to identify "hot" areas during endoscopy.
• Assessment of AI performance in real-time clinical practice.

Main Results:

• Deep learning AI systems demonstrate high accuracy in retrospective image analysis.
• Real-time AI testing in randomized trials addresses clinical practice pitfalls like poor bowel preparation.
• Emerging data suggests AI's potential to improve polyp detection rates.

Conclusions:

• AI systems offer a promising approach to enhance colonoscopy quality.
• Real-time AI application may significantly reduce missed colorectal neoplasia and interval CRC.
• Further research and clinical integration of AI are warranted to optimize colorectal cancer screening.