Masterclass: Advanced GI Endoscopic Imaging (On-Demand) | December 2025-Computer Aided Detection and Diagnosis in Barrett’s Esophagus

Video Transcription

Video Summary

Professor Jacques Bergman from the Netherlands discusses advances in artificial intelligence (AI) for enhancing Barrett's esophagus endoscopy. His team pioneered AI tools to detect and diagnose early neoplasia, using machine and deep learning models trained on specialized image datasets. Unlike colonoscopy, Barrett's lesions are subtler and less numerous, making AI development challenging. Their international Bonsai consortium created AI algorithms outperforming general endoscopists and matching experts, especially when AI assists clinicians. However, AI struggles with variable image quality typical in community settings, causing a "domain gap" between academic and real-world applications. Strategies to close this gap include more robust algorithms trained on diverse endoscopic images and quality control systems ensuring high-quality input. Bergman emphasizes the importance of optimizing human-AI interaction to build trust and improve detection. For regulatory approval, he suggests benchmarking AI's ability to enhance general endoscopists’ detection rather than large clinical trials, and ensuring tools don’t disrupt workflow. AI shows promise but requires robust, practical integration and streamlined regulation for clinical success.

Asset Subtitle

Jacques Bergman

Meta Tag

Concept Barretts Esophagus

Concept Endoscopy

Concept Computer-Aided Detection

Concept Computer-aided Diagnosis

Concept Domain Shift

Keywords

Barrett's esophagus

artificial intelligence

endoscopy

early neoplasia detection

machine learning

clinical integration

Barretts Esophagus

Endoscopy

Computer-Aided Detection

Computer-aided Diagnosis

Domain Shift