Population-Scale Deep Learning & Large-Scale Imaging Genetics

Detecting subtle imaging biomarkers at an early disease stage requires deep learning models trained on large, unselected, and highly representative population cohorts.

Research Focus:

• • Leveraging multi-modal imaging data from large cohorts such as the German National Cohort (NAKO Gesundheitsstudie) to build robust and highly generalizable deep learning architectures.

  • Quantitative phenotyping and discovery of novel imaging biomarkers for neurodegenerative and vascular conditions.

  • Application of transfer learning strategies from population-level data to target clinical cohorts to prevent overfitting and algorithmic bias.