AbdomenAtlas

The largest annotated abdominal CT dataset: 20K+ volumes from 112 hospitals with per-voxel organ annotations built by an AI-accelerated human-in-the-loop engine (Medical Image Analysis 2024; ICLR 2024 oral lineage), plus the SuPreM suite of pretrained 3D segmentation models with demonstrated transfer to new targets. JHU CCVL (Wenxuan Li, Zongwei Zhou, Alan Yuille) — an on-thesis medical foundation-model data engine.