EAGLE Trial Demonstrates Olympus CADDIE AI Enhances Detection of High-Risk Colorectal Lesions

The EAGLE Trial, a multicenter randomized controlled study evaluating the Olympus CADDIE cloud-based Computer-Aided Detection application, demonstrates that artificial intelligence can significantly enhance the detection of high-risk colorectal lesions during colonoscopy. Published in npj Digital Medicine, the study involved 841 patients across eight centers in four European countries and found that CADDIE-assisted colonoscopy increased adenoma detection rate by 7.3% compared to standard procedures.

Clinically significant improvements were observed for specific lesion subtypes that are critical for colorectal cancer prevention. The study reported a 93% relative increase in detection of large adenomas (>10 mm), a 57% increase for non-polypoid adenomas, and a remarkable 230% increase for sessile serrated lesions (SSLs). These flat and sessile lesions are particularly challenging to detect visually but represent high-risk precursors to colorectal cancer. Recent research, including studies published in Gastroenterology and Lancet Gastroenterology & Hepatology, has established that SSL detection is crucial for reducing post-colonoscopy cancer risk.

The CADDIE application represents the first cloud-based CADe solution with both FDA clearance and CE marking, operating as part of the OLYSENSE Intelligent Endoscopy Ecosystem. Cloud deployment offers practical advantages for healthcare institutions, including reduced hardware dependency and subscription-based access models that could democratize advanced AI tools. The system demonstrated real-time performance across diverse clinical environments without disrupting workflow efficiency or safety protocols.

Principal Investigator Rawen Kader noted that cloud deployment removes hardware barriers and provides hospitals with access to the latest AI innovations. The study addresses concerns raised in recent guidelines from the European Society of Gastrointestinal Endoscopy and the American Gastroenterological Association regarding AI-assisted colonoscopy. The complete study is available at https://doi.org/10.1038/s41746-025-02270-1.

Executive leadership emphasized the trial's significance for clinical translation of AI in routine endoscopy. The CADDIE application is specifically trained on datasets enriched with clinically relevant and hard-to-detect lesions, aligning with quality indicators for colonoscopy that increasingly prioritize reliable SSL detection. This evidence-based approach supports the integration of AI tools that enhance detection of lesions most relevant to cancer prevention while maintaining procedural efficiency and safety standards.