AbdomenAtlas
收藏资源简介:
AbdomenAtlas是由约翰斯·霍普金斯大学领衔创建的大型腹部CT数据集,包含20,460个三维CT体积,来自112家医院,覆盖多样的地理和人口。该数据集提供了673,000个高质量的解剖结构掩码,由10名放射科医生借助AI算法标注。数据集的创建过程包括专家手动标注和半自动标注,旨在支持大规模AI模型的开发和算法基准测试。AbdomenAtlas的应用领域广泛,特别是在医疗图像分析中,旨在提高AI算法在复杂临床场景中的性能和可靠性。
AbdomenAtlas is a large-scale abdominal CT dataset led by Johns Hopkins University. It contains 20,460 3D CT volumes collected from 112 hospitals, covering diverse geographic and demographic backgrounds. The dataset provides 673,000 high-quality anatomical structure masks, which were annotated by 10 radiologists with the aid of AI algorithms. Its development process integrates expert manual annotation and semi-automatic annotation, and is designed to support the development of large-scale AI models and algorithm benchmarking. AbdomenAtlas has a wide range of application scenarios, particularly in medical image analysis, with the objective of enhancing the performance and reliability of AI algorithms in complex clinical settings.




