The EAGLE Trial, a multicenter randomized controlled study evaluating the CADDIE™ device, represents a significant advancement in colorectal cancer prevention through artificial intelligence-assisted endoscopy. Published in npj Digital Medicine, this study demonstrates that cloud-deployed AI can help endoscopists detect the lesions that matter most in preventing progression to cancer—large adenomas, particularly those flat in morphology, and sessile serrated lesions (SSLs)—without disrupting safety or workflow. The trial, conducted across eight centers in four European countries, involved 841 patients and 22 endoscopists performing screening and surveillance colonoscopies, with patients randomized to standard colonoscopy or CADDIE-assisted colonoscopy.
In screening and surveillance patients, use of the CADDIE™ application was associated with a 7.3% absolute increase in adenoma detection rate compared to standard colonoscopy. More importantly, the study showed significant relative increases in lesions detected per colonoscopy for clinically relevant lesion subtypes: 93% for large (>10 mm) adenomas, 57% for non-polypoid adenomas, and 230% for SSLs. These findings are particularly significant because lesions with sessile or flat morphology are difficult to detect and can harbor clinically relevant pathology. SSLs, in particular, are high-risk lesions whose detection is critical to reducing the risk of post-colonoscopy colorectal cancer, as noted in studies published in Gastrointestinal Endoscopy and The Lancet Gastroenterology & Hepatology.
The CADDIE™ application is trained on a dataset enriched in clinically relevant and hard-to-detect lesions, including flat sessile serrated lesions and large polyps. This study demonstrates increased detection of clinically relevant lesions and no increase in unnecessary resections, addressing some of the concerns raised in recent guidelines from the European Society of Gastrointestinal Endoscopy and the American Gastroenterological Association. The ability to reliably detect SSLs is increasingly viewed as a critical quality consideration in colonoscopy, as highlighted in quality indicators published in Gastrointestinal Endoscopy.
The cloud architecture of the CADDIE™ application represents a key innovation, using industry standard security controls while offering hospitals flexibility, reducing reliance on hardware, and enabling subscription-based procurement models. This approach can democratize access to advanced AI tools and lays the foundation for future AI applications in endoscopy. Rawen Kader, Principal Investigator of the EAGLE Trial and GI Researcher at University College London, noted that "cloud deployment can remove hardware barriers and give hospitals access to the latest AI innovations, which has the potential of improving detection of the lesions that matter most for reducing colorectal cancer risk."
The complete study findings are available at https://doi.org/10.1038/s41746-025-02270-1. The CADDIE computer-assisted detection device is limited for use with standard white-light endoscopy imaging only and is not intended to replace a full patient evaluation or be relied upon to make a primary interpretation of endoscopic procedures, medical diagnosis, or treatment recommendations.
