Dataset Hub

AbdomenAtlas 1.0 Mini is a large medical‑image segmentation dataset containing 5,195 annotated CT volumes, used for segmenting multiple organs including spleen, liver, kidney, stomach, gallbladder, pancreas, aorta, and inferior vena cava. The dataset is maintained by the CCVL research group at Johns Hopkins University and is intended for comparing and evaluating various semantic‑segmentation and pre‑training algorithms.

SA-Med2D-20M是一个大规模的2D医学图像分割数据集，包含4.6百万张2D医学图像和19.7百万个对应的掩码，覆盖了几乎整个身体，并展示了显著的多样性。该数据集旨在帮助研究人员构建医学视觉基础模型或将其模型应用于下游医学应用。