Vision-Language Foundation Models for Cardiac Imaging

Explore the development and application of vision-language foundation models specifically designed for cardiac imaging analysis in this 37-minute conference presentation. Learn about EchoPrime, a groundbreaking multi-video vision-language model that can generate comprehensive reports for echocardiogram studies, mimicking the holistic interpretation approach of cardiologists. Discover the technical challenges and methodologies involved in building large multimodal models (LMMs) that synthesize information from multiple cardiac imaging sources rather than focusing on single, task-specific applications. Examine how these advanced AI systems can potentially transform clinical practice by automating and enhancing various aspects of the cardiac imaging pipeline. Gain insights into the intersection of machine learning, bioengineering, and cardiology from a researcher who specializes in training large multimodal models and developing ML-powered solutions for medical imaging applications.