AMOS

AMOS是一个大规模的腹部多器官分割基准数据集，由香港大学深圳大数据研究院创建。该数据集包含500例CT和100例MRI扫描，涵盖了多中心、多供应商、多模态、多阶段、多疾病的患者数据，每例数据都有15个腹部器官的体素级标注。AMOS数据集旨在为研究鲁棒分割算法提供挑战性的示例和测试平台，适用于多种目标和场景。此外，AMOS还可用于多种学习任务，如分布外泛化、跨模态学习、迁移学习等，为医学图像分析领域提供了丰富的资源。

AMOS is a large-scale abdominal multi-organ segmentation benchmark dataset developed by the Shenzhen Institute of Big Data, The University of Hong Kong. This dataset comprises 500 CT scans and 100 MRI scans, covering patient data from multiple centers, vendors, modalities, scan phases and disease states, with voxel-level annotations for 15 abdominal organs per case. The AMOS dataset aims to provide challenging samples and a testbed for research on robust segmentation algorithms, suitable for various research objectives and application scenarios. Furthermore, AMOS can be applied to multiple learning tasks including out-of-distribution generalization, cross-modal learning, transfer learning and more, serving as a rich resource for the field of medical image analysis.