Unified Trustworthy Multimodal Foundation Model for Radiology

This project aims to build a unified multimodal foundation model for radiology that integrates imaging, textual reports, time series data, and demographics to enhance diagnostic accuracy, report generation, and medical education. The team is also pioneering uncertainty quantification for trustworthy AI deployment in hospitals. Early efforts focus on mammography and chest X-rays using data from NYU Langone and public ICU datasets.