New Member Double-Header

Zoom information will be sent to e-HAIL members.

Beyond the Black Box: Generative AI and Synthetic Data for Trustworthy Medical Image Analysis – Matthias Wilms, PhD
Generative AI tools such as ChatGPT are transforming multiple aspects of our lives, promising to help us achieve things faster and better. These models are also profoundly impacting medical image analysis. However, a critical barrier prevents the clinical deployment of many AI systems: a lack of trustworthiness. Clinicians and patients need to understand how and why AI systems make their predictions, and we need to ensure that these systems generalize well. In my talk, I will discuss how generative models and synthetic imaging data generated by them can address these challenges in medical image analysis. First, I will present our recent work using generative models as self-explanatory classifiers that can visually show their reasoning process rather than operating as black boxes. Second, I will demonstrate how synthetic imaging data enables systematic analyses of algorithmic biases, which help us to understand when and why deep learning models may fail.

Advancing Healthcare Through AI and Machine Learning Innovations – Joyce Yan-Ran Wang, PhD
Healthcare accounts for roughly 10% of the global economy, yet it remains riddled with challenges such as inequitable access, an unprecedented demographic shift toward older populations, and unsustainably rising per capita costs. This convergence of forces signals an urgent need for innovative solutions. In my talk, I will explore how AI innovations, driven by the pressing needs of healthcare, are advancing both clinical and technological frontiers. I will present the impactful work of our team in developing safer medical imaging, automated diagnosis for cardiovascular diseases, and AI-powered precision oncology. Central to this work are specialized deep learning algorithms specifically designed for healthcare, with careful consideration of the underlying biological complexities.