A report-grounded vision-language foundation model for colonoscopy from 280000 routine reports

Jia Yu, Yan Zhu, Yili He, Zilong Wang, Xinyang Jiang et al.|July 30, 2026arXiv

Key Takeaway

Routine clinical reports contain rich supervision signals—you can extract finding-to-frame correspondence at scale to train medical vision-language models without expensive frame-level annotation, turning existing documentation into training data.

Summary

EndoCLIP is a vision-language model trained on 280,000 colonoscopy reports to link clinical findings with individual video frames.

multimodal data

Key Terms

vision-language-models zero-shot-learning linear-probes weak-supervision contrastive-learning