COVID-CT-Rate.zip

COVID-CT-Rate is a dataset including 433 CT images from 82 COVID-19 patients with their associated infection masks. It can be used for training AI models to segment COVID-19 lesions from chest CT images.For the annotation process, first, infection masks were generated using a standard U-Net pre-trained on a public COVID-19 dataset. Then, a thoracic radiologist with 20 years of experience in lung imaging carefully modified and verified the generated infection masks. All CT images have been obtained without contrast enhancement and saved in the Digital Imaging and Communications in Medicine (DICOM) format and the Hounsfield Unit.CT images have been selected from diffident parts of the lung (top, middle, and bottom) with different infection rates to help the AI model better predict the infection regions on unseen CT images from the whole lung volume.DICOM files for each patient have been saved in individual folders within the COVID-Rate-CT directory, and corresponding infection masks, labeled with patient and image numbers, have been stored in the COVID-Rate-Masks directory.For ease of use, you can also download CT images and infection masks from the <code>CTs.npy and </code>InfMasks<code>.npy</code> files. <br>If you found this dataset helpful to your research, please consider citing:<br>Enshaei N, Oikonomou A, Rafiee MJ, Afshar P, Heidarian S, Mohammadi A, Plataniotis KN, Naderkhani F. COVID-rate: an automated framework for segmentation of COVID-19 lesions from chest CT images. Scientific Reports. 2022 Feb 25;12(1):3212.

COVID-CT-Rate是一款涵盖82名新冠患者的433张胸部CT图像及其对应感染区域掩码的数据集，可用于训练AI模型以从胸部CT图像中分割新冠病变区域。在标注流程中，研究人员首先使用在公开新冠数据集上预训练的标准U-Net生成感染区域掩码，随后由一名拥有20年肺部影像阅片经验的胸部放射科医师对生成的掩码进行细致修改与验证。所有CT图像均未使用对比增强扫描获取，以医学数字成像和通信（Digital Imaging and Communications in Medicine, DICOM）格式及霍恩斯菲尔德单位（Hounsfield Unit, HU）存储。本次数据集选取了肺部不同区域（肺尖、肺中部、肺底部）且感染率各异的CT图像，旨在助力AI模型更好地对全肺体积的未知CT图像进行感染区域预测。每位患者的DICOM文件均存储于COVID-Rate-CT目录下的独立子文件夹中，对应的感染区域掩码（以患者及图像编号命名）则存放于COVID-Rate-Masks目录下。为便于使用，您也可通过<code>CTs.npy</code>与<code>InfMasks.npy</code>文件下载CT图像及感染区域掩码。<br>若本数据集对您的研究有所助益，请考虑引用以下文献：<br>Enshaei N, Oikonomou A, Rafiee MJ, Afshar P, Heidarian S, Mohammadi A, Plataniotis KN, Naderkhani F. COVID-rate: an automated framework for segmentation of COVID-19 lesions from chest CT images. Scientific Reports. 2022 Feb 25;12(1):3212.