MedVTok: A General-Purpose Medical Visual Tokenizer
One visual language for all medical scans, regardless of type or dimension
Researchers created MedVTok, a single system that converts medical images—from X-rays and ultrasounds to 3D CT scans—into a unified format that AI systems can understand. Trained on 28 million images across nine different imaging types, the system outperformed existing methods on 24 different medical tasks, from diagnosis to report writing, all using the same underlying representation.
Hospitals use dozens of incompatible imaging systems (MRI, CT, ultrasound, X-ray), forcing doctors and AI systems to translate between formats and forcing developers to build separate tools for each. A unified medical image language cuts development time and makes AI tools portable across hospitals—one trained model works on any scan type, reducing both costs and the time needed to deploy AI diagnostics in new settings.