How AI is Transforming Colorectal Cancer Screening

By Campion Quinn, MD

Colorectal cancer (CRC) is a leading cause of cancer deaths worldwide. Colonoscopy remains the gold standard for preventing CRC by detecting and removing polyps. However, up to 25% of polyps can be missed during a standard colonoscopy, leading to interval cancers that develop between scheduled screenings​(AI colonoscopy 1)​(ai in colonoscopy 2). Artificial intelligence (AI), specifically through Computer-Aided Detection (CADe) systems, is now helping to reduce these missed polyps, improving adenoma detection rates (ADR) and reducing the adenoma miss rate (AMR). Here’s how AI is changing the game for both physicians and patients.

Reducing Missed Polyps with AI

AI-assisted colonoscopy has shown impressive results in reducing missed polyps. Studies show that AI can cut the adenoma miss rate (AMR) from 32.4% to 15.5%​(AI colonoscopy 1). This reduction is particularly important because even small, subtle lesions that go undetected can eventually progress to colorectal cancer. By acting as a second set of eyes, AI helps catch these polyps, especially the smaller ones (≤5 mm) and flat lesions that are more difficult for the human eye to detect​(ai in colonoscopy 2).

Boosting Adenoma Detection Rates

The adenoma detection rate (ADR) is a key measure of a colonoscopy’s effectiveness. Higher ADRs are linked to lower rates of colorectal cancer. AI has been shown to improve ADR significantly. In a meta-analysis, AI-assisted colonoscopies achieved an ADR of 36.6%, compared to 25.2% with standard colonoscopies​(ai in colonoscopy 2). This improvement is mainly due to AI’s ability to detect small, often overlooked polyps, which can still pose a risk for cancer development.

Real-World Impact: Tandem Colonoscopy Studies

In real-world clinical settings, tandem colonoscopy studies—where patients undergo two colonoscopies back-to-back, one with AI and one without—have confirmed AI’s effectiveness. In one study, the adenoma miss rate was reduced from 40% in the standard group to just 13% in the AI group​(AI in colonoscopy 2). This significant reduction shows how AI can improve the thoroughness of colonoscopies and reduce the chances of missed lesions.

Practical Considerations for Physicians

While AI offers many benefits, it’s essential to know its limitations. When AI flags a normal area as suspicious, false positives can slow down the procedure. However, most false positives are quickly dismissed by experienced endoscopists, making their impact minimal​(ai in colonoscopy 2). Another consideration is the quality of the AI system itself. Systems like GI Genius and EndoBRAIN are already making strides in improving detection rates, but the effectiveness of any AI tool depends on the data it was trained on. Choosing an AI system trained on a diverse set of images can help ensure better performance in various clinical situations.

Conclusion

AI is transforming colorectal cancer screening by improving detection rates and reducing the risk of missed polyps. For practicing physicians, incorporating AI into routine colonoscopies can lead to better patient outcomes and potentially prevent more cases of colorectal cancer. As AI technology continues to advance, it’s clear that it will play an increasingly vital role in colonoscopy and cancer prevention.