HR-Kidney

HR-Kidney数据集由苏黎世大学生理学研究所创建，是目前最大的监督式3D生物医学研究图像训练数据集，包含33,729个肾小球的高分辨率3D图像，数据量达1.7TB。该数据集基于同步辐射X射线相位对比微断层摄影技术，提供了完整的鼠肾图像及验证过的肾小球分割。创建过程中，利用了先进的成像设施和专业的领域知识进行标注。该数据集广泛应用于图像处理、数据增强和机器学习等领域，特别适用于无监督和半监督学习研究，以及迁移学习和生成对抗网络的开发。

The HR-Kidney dataset was created by the Institute of Physiology, University of Zurich. It is currently the largest supervised 3D biomedical imaging training dataset for scientific research, containing high-resolution 3D images of 33,729 glomeruli with a total data volume of 1.7 TB. Based on synchrotron radiation X-ray phase-contrast microtomography technology, this dataset provides complete mouse kidney images along with validated glomerular segmentations. Advanced imaging facilities and specialized domain expertise were employed for annotation during its development. Widely applied in fields such as image processing, data augmentation and machine learning, this dataset is particularly suited for unsupervised and semi-supervised learning research, as well as the development of transfer learning and generative adversarial networks (GANs).