A trial of deep-learning detection in colonoscopy

A double-blind randomized trial in China has assessed the effect of operational bias in the use of computer-aided detection (CAD) systems for polyp detection. A total of 962 patients presenting for diagnostic and screening colonoscopy were randomly allocated to colonoscopy with a CAD or a sham system; both patient and operator were unaware of the assigned group. The adenoma detection rate was significantly greater in the CAD group than the sham group (P = 0.03). Polyps detected by the CAD system but not the endoscopist had difficult-to-recognise characteristics (small, flat, isochromatic and unclear boundaries).


Original article

  1. Wan, P. et al. Effect of a deep-learning computer-aided detection system on adenoma detection during colonoscopy (CADe-DB trial): a double-blind randomised study. Lancet Gastroenterol. Hepatol. (2020)

    Article  PubMed  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Iain Dickson.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Dickson, I. A trial of deep-learning detection in colonoscopy. Nat Rev Gastroenterol Hepatol 17, 194 (2020).

Download citation


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing