Talk from Andras Jakab - University of Zurich
Short abstract:
Fetal and neonatal MRI offers a unique window into early human brain development but suffers from severe motion, low SNR, and limited resolution. In this talk, I will introduce how our team integrates advanced post-processing and AI-based tools, such as domain-adapted segmentation or generative AI, to recover anatomically faithful, quantitative datasets. We extend these methods to connectivity analysis using graph learning and deep regression models for brain-age estimation. Applied to large cohorts of fetuses and infants with congenital heart defects, spina bifida, and prematurity, these techniques reveal disease-specific developmental trajectories. This talk will highlight how computational imaging and AI are transforming early-life MRI from qualitative visualization to quantitative neurophenotyping.