Our paper "Predicting Longitudinal Brain Development via Implicit Neural Representations” presents the first atlas capable of forecasting individualized fetal and neonatal brain trajectories from MRI, moving beyond traditional atlases that only capture generic population trends. Our model predicts subject-specific development up to ±20 weeks and incorporates external factors such as birth age and birth weight. This conditioning enables realistic "what-if” simulations of neurodevelopment, providing a novel tool for studying personalized brain maturation and the influence of perinatal risk factors.
It is a privilege to present at MICCAI 2025, the leading venue in medical image computing. Sharing my work with this community is both motivating and inspiring, and it is an important opportunity to receive feedback from experts across a wide variety of fields. I believe their insight will help shape the next steps of this research.
I am particularly excited to learn about advances in generative models, longitudinal analysis, and applications of AI to clinical neuroimaging. I am also interested in methods that improve data efficiency and interpretability, which are crucial when working with rare or sensitive medical data.
Beyond the scientific program, I look forward to connecting with colleagues from around the world, exchanging ideas, and exploring possible collaborations. I am also excited to experience Daejeon and the unique cultural and social events that MICCAI always provides.