Description of this dataset

Dataset of the chest X-Ray images (CXRs) is constructed from cohorts from the China Consortium of Chest X-ray Image Investigation (CC-CXRI). All CXRs are classified into COVID-19 pneumonia due to SARS-CoV-2 virus infection, other viral pneumonia, bacterial pneumonia, other lung disorders, and normal controls.

ChestDx and ChestDx-PE were used to developed AI models for identifying common chest diseases with labels of 14 common thoracic pathologies, including atelectasis, cardiomegaly, consolidation, edema, effusion, emphysema, fibrosis, hernia, infiltration, nodule, mass, pleural thickening, pneumonia, and pneumothorax. ChestDx is a dataset which consists of patients from hospital visits; ChestDx-PE is another dataset which consists of additional patients who underwent a routine annual physician examination.

For pneumonia diagnosing/triaging with an application to COVID-19 pneumonia, the dataset (CC-CXRI-P) consists of viral pneumonia (including COVID-19 pneumonia), other types of pneumonia, and normal controls.

This dataset and AI code are available globally with the aim to assist the clinicians and researchers to combat the COVID-19 pandemic.

Please cite this paper

Guangyu Wang, Xiaohong Liu, Jun Shen, et al. Kang Zhang, Weimin Li, Tianxin Lin. (2021). A deep-learning pipeline for the diagnosis and discrimination of viral, non-viral and COVID-19 pneumonia from chest X-ray images. Nature Biomedical Engineering, https://dx.doi.org/10.1038/s41551-021-00704-1