Generative AI for Neonatal Medical Imaging

by Sayane Shome

Tuesday 29 Jul 2025, 12:30 → 13:30 America/Los_Angeles

48/2-224 - Madrone (SLAC)

Generative AI is rapidly transforming medical imaging, opening new opportunities for research, clinical decision-making, and education. The talk will begin with an overview of foundational generative models—such as GANs and diffusion models—and highlight key state-of-the-art applications currently shaping the medical imaging landscape. It will then focus on ongoing work leveraging diffusion-based models for critical neonatal applications, including gestational age prediction, synthesizing neonatal chest X-rays from clinical text prompts, and incorporating patient-specific details into image generation.

 

These capabilities underscore the potential of generative AI to create synthetic datasets, enhance diagnostic support, and enable research in low-data clinical settings. Inherent challenges inherent to real-world neonatal datasets will also be discussed, along with segmentation-based strategies for mitigating these issues. The session will conclude with future directions for integrating these methods into clinical workflows and their broader implications for applying generative AI beyond medicine.

 

https://stanford.zoom.us/j/98973156241?pwd=cEU5RFdlVXoyc0JTeTlDMkozKzQ5UT09

slac_presentation_july29_2025.pptx