Radiology, long at the forefront of AI adoption in healthcare, is undergoing a profound transformation. The shift from convolutional neural networks (CNNs) to foundation models and vision‑language models (VLMs) is redefining how we extract insights from imaging data and integrate them with clinical narratives. This talk explores the potential of these multimodal AI advancements to revolutionize radiology and healthcare at large.

We will discuss how foundation models, capable of generalizing across tasks, and VLMs, bridging imaging and text, are enabling:

• Advanced image analysis beyond classification, enabling contextual understanding and natural language queries.
• Integration of imaging data with electronic health records and reports for richer diagnostic insights.
• Enhanced workflows for radiologists, transforming them into orchestrators of AI‑assisted decision-making.
• Through real-world examples, we will demonstrate how these models surpass traditional CNN‑based systems in robustness, scalability, and adaptability. We’ll also address challenges such as explainability, fairness, and the need for rigorous validation in clinical settings.

Radiology stands as a microcosm for the broader healthcare AI revolution. This session invites you to explore how foundation models and VLMs are not just tools for imaging but catalysts for a new era of multimodal intelligence in healthcare.