Train-free unsupervised volumetric embeddings using image foundation models enable improved predictions and genomic discoveries
RECOMB Genetics, 2025
*Equal contribution
A train-free approach to volumetric representation learning for medical prediction and imaging-genetics applications.