人体正常肺部CT图像数据集

一个公开的开放数据集，包含COVID-19阳性或疑似患者的胸部CT图像、其他病毒性肺炎以及正常肺部CT图像。数据来源于公共资源和医院及医生的间接收集，所有图像和数据将在此GitHub仓库中公开发布。本数据汇交技术方案所涉及的联邦学习数据按照课题指标形成4个文件夹，即一级文件夹/2021ZD0113904-Covid19-Datasets,/2021ZD0113904-Covid19-Datasets-Client1,/2021ZD0113904-Covid19-Datasets-Client2,/2021ZD0113904-Covid19-Datasets-Client3。各数据集文件夹包括3个子文件夹，即二级子文件夹包括训练数据集，测试数据集，验证数据集，即/train，/test，/val。图表合集子文件夹内有若干图表，支撑数据子文件夹内有若干原始数据，即三级子文件夹包括/Normal。数据集文件夹命名格式为：课题编号-Covid19-数据集名称，此为一级文件夹。课题编号-Covid19-数据集-训练类别，此为二级文件夹。课题编号-Covid19-数据集-训练类别-数据类型，此为三级文件夹。数据集内图文件命名格式为:课题编号-Covid19-数据集-训练类别-数据类型-图x，在图合集文件内，只保留“图x”。图片格式为.jpeg和.png。

This is a public open dataset containing chest CT images of COVID-19-positive or suspected patients, other viral pneumonia cases, and normal lung CT images. The data is sourced from public resources and indirect collections from hospitals and physicians, and all images and data will be publicly released in this GitHub repository. The federated learning data involved in this data submission technical plan is divided into 4 folders according to the project indicators, namely the primary folders: /2021ZD0113904-Covid19-Datasets, /2021ZD0113904-Covid19-Datasets-Client1, /2021ZD0113904-Covid19-Datasets-Client2, /2021ZD0113904-Covid19-Datasets-Client3. Each dataset folder includes 3 secondary subfolders, namely the training dataset, test dataset, and validation dataset, i.e., /train, /test, /val. There are several charts in the chart collection subfolder and several raw data in the supporting data subfolder, specifically the tertiary subfolders including /Normal. The naming convention for dataset folders is: Project ID-Covid19-Dataset Name, which serves as the primary folder. The naming convention for secondary folders is: Project ID-Covid19-Dataset-Training Category. The naming convention for tertiary folders is: Project ID-Covid19-Dataset-Training Category-Data Type. The naming format for image files in the dataset is: Project ID-Covid19-Dataset-Training Category-Data Type-Image x, where only "Image x" is retained in the chart collection files. The supported image formats are .jpeg and .png.