What Is Computer-Aided Polyp Detection and Should Gastroenterologists Use It?

Answer in brief: Computer-aided detection (CADe) uses artificial intelligence to analyze live colonoscopy video and alert the endoscopist to possible polyps. Randomized trials show that CADe can increase adenoma detection rate, largely by identifying additional small lesions. However, the 2025 AGA guideline made no recommendation for or against routine CADe-assisted colonoscopy because evidence is still very uncertain for the outcomes that matter most: colorectal cancer incidence, colorectal cancer mortality and post-colonoscopy colorectal cancer. CADe is best viewed as an adjunct to, not a substitute for, high-quality inspection.

What CADe actually does

CADe systems process the video stream in real time and place a visual or audible marker around an area that may represent a polyp. They do not independently diagnose histology and do not remove lesions. A separate category, computer-aided diagnosis, attempts to characterize lesions. In routine practice the most mature application is detection support.

Why adenoma detection improves

Human attention is imperfect, particularly for subtle or briefly visible lesions. CADe acts as a persistent second observer and increases adenoma detection in many randomized studies. The absolute gain varies with baseline endoscopist performance, technology, population and study setting. Some of the added lesions are diminutive, which raises questions about clinical significance and downstream surveillance burden.

Why AGA did not recommend routine adoption

The 2025 AGA living guideline judged long-term evidence to be of very low certainty. Improved adenoma detection is an important surrogate, but it is not yet proven that CADe lowers cancer incidence, cancer mortality or post-colonoscopy cancer. Cost, workflow, false-positive prompts, operator dependence and how additional polyp detection affects surveillance intervals also remain relevant.

How to use it responsibly

If CADe is used, the program should monitor standard colonoscopy quality indicators rather than treating AI adoption itself as proof of quality. Adequate bowel preparation, cecal intubation, careful mucosal exposure, sufficient withdrawal inspection, complete resection and appropriate surveillance remain foundational. AI should enhance vigilance without encouraging faster withdrawal or cognitive offloading.

Practical clinical algorithm

Evaluate whether the technology is validated for the local endoscopy platform and population.
Introduce CADe as an adjunct to standard high-quality colonoscopy, not a replacement.
Track ADR, withdrawal quality, false-positive burden and downstream surveillance effects.
Continue standard training and quality improvement regardless of AI use.
Reassess adoption as living guideline evidence matures.

Common mistakes to avoid

Assuming higher ADR automatically means fewer cancers.
Letting CADe replace careful mucosal exposure and inspection.
Using AI adoption as a marketing quality metric without outcome monitoring.
Ignoring the surveillance burden created by detecting more diminutive lesions.

Trainee takeaway

CADe can help endoscopists see more polyps. Whether that translates into fewer colorectal cancers is not yet established, which is why current AGA guidance remains neutral.

Frequently asked questions

Does CADe increase adenoma detection?

Yes, on average in randomized studies.

Does CADe reduce colorectal cancer mortality?

That has not yet been established.

Should every endoscopy unit buy CADe now?

Current AGA guidance makes no recommendation for or against routine use; local quality, cost and workflow matter.

Relevant free books from Dr. Thomson

Endoscopy and Diagnostic Imaging — available as a free digital download from GIandHepatology.com.
Images in Gastroenterology and Hepatology — available as a free digital download from GIandHepatology.com.

References

1. Thomson ABR. Endoscopy and Diagnostic Imaging. CAPstone Academic Publishers.

2. Sultan S, Shung DL, Kolb JM, et al. AGA Living Clinical Practice Guideline on Computer-Aided Detection-Assisted Colonoscopy. Gastroenterology. 2025. doi:10.1053/j.gastro.2025.01.002.

3. AGA. Use of computer-aided detection systems (CADe) in colonoscopy. Published March 20, 2025.

Educational content only. Clinical decisions should incorporate the individual patient, local resources, product labeling, and the most current applicable guideline.