Introduction
Artificial intelligence (AI) is playing an increasingly prominent role in gastrointestinal endoscopy, particularly in enhancing real-time lesion detection during colonoscopy. This study evaluates the impact of an advanced AI-assisted detection system on the Adenoma Detection Rate (ADR), Adenomas per Colonoscopy (APC), and predictive performance, comparing outcomes across endoscopists with different levels of experience.
Aims & Methods
Three endoscopists participated in the study: two with limited experience (Physicians A and B) and one expert (Physician C). A total of 720 colonoscopies were analyzed, with 400 procedures performed using AI assistance and 320 without, evenly distributed among the three operators.
Patients included were between 40 and 75 years of age and underwent colonoscopy either as a screening examination or following a positive fecal occult blood test. Exclusion criteria comprised a history of inflammatory bowel disease (IBD), hereditary polyposis syndromes, major colorectal surgery, or inadequate bowel preparation. For each procedure, data were collected on ADR, APC, the predictive accuracy of the AI system, and retraction time (from cecum to anus).
Results
The use of AI resulted in a marked and consistent improvement in all primary performance indicators across all three endoscopists.
In terms of ADR, Physician A improved from 23.8% without AI to 29.7% with AI; Physician B from 26.9% to 31.6%; and Physician C from 34.7% to 36.6%. Similarly, the APC index increased from 0.52 to 0.63 for Physician A, from 0.55 to 0.65 for Physician B, and from 0.64 to 0.73 for Physician C.
The AI system demonstrated high predictive accuracy across all operators, with no false positives recorded for polyps ≥4 mm in size. Prediction rates were 87.9% for Physician A, 89% for Physician B, and 92% for Physician C, confirming the device’s reliability in supporting polyp identification during real-time examination. Retraction times were slightly prolonged with AI usage: 8.3 minutes vs 7.5 for Physician A, 8.7 vs 7.4 for Physician B, and 8.8 vs 8.0 for Physician C. However, these differences did not reach statistical significance, indicating that AI support did not meaningfully affect procedural efficiency.
| Physician | ADR (%) – No AI
| ADR (%) – With AI
| APC – No AI
| APC – With AI
|
| A | 23.8 | 29.7 | 0.52 | 0.63 |
| B | 26.9 | 31.6 | 0.55 | 0.65 |
| C | 34.7 | 36.6 | 0.64 | 0.73 |
Conclusion
The integration of AI into colonoscopy workflows significantly improved both ADR and APC, with the most substantial impact observed in less experienced endoscopists. Physicians A and B, who initially showed suboptimal detection metrics, achieved levels of quality comparable to those of the expert operator in the absence of AI, highlighting AI’s value as a tool for bridging the gap in operator performance. The high predictive reliability of the system—particularly for clinically relevant lesions—and the absence of false positives ≥4 mm further support its clinical utility. Although retraction times increased slightly with AI, this change was not statistically significant, suggesting that the diagnostic advantages were not offset by decreased procedural efficiency. Overall, AI proves to be a valuable and scalable solution to enhance adenoma detection, support clinical training, and promote uniform quality standards in colorectal cancer screening.
References
1.Antonelli, G., Gkolfakis, P., Tziatzios, G., Papanikolaou, I. S., Triantafyllou, K., & Hassan, C. (2020). Artificial intelligence-aided colonoscopy: Recent developments and future perspectives. World journal of gastroenterology, 26(47), 7436–7443.
2.Hassan, C., Balsamo, G., Lorenzetti, R., Zullo, A., & Antonelli, G. (2022). Artificial Intelligence Allows Leaving-In-Situ Colorectal Polyps. Clinical gastroenterology and hepatology : the official clinical practice journal of the American Gastroenterological Association, 20(11), 2505–2513.e4
3.Houwen, B. B. S. L., Hassan, C., Coupé, V. M. H., Greuter, M. J. E., Hazewinkel, Y., Vleugels, J. L. A., Antonelli, G., Bustamante-Balén, M., Coron, E., Cortas, G. A., Dinis-Ribeiro, M., Dobru, D. E., East, J. E., Iacucci, M., Jover, R., Kuvaev, R., Neumann, H., Pellisé, M., Puig, I., Rutter, M. D., … Dekker, E. (2022). Definition of competence standards for optical diagnosis of diminutive colorectal polyps: European Society of Gastrointestinal Endoscopy (ESGE) Position Statement. Endoscopy, 54(1), 88–99.
