COVID-19 imaging datasets

Overview of open access COVID-19 imaging datasets for teaching, training and research

Imaging databases and registries are essential for diagnosis in radiology, as well as for the development of artificial intelligence tools for machine-based diagnosis. EIBIR has compiled a preliminary list of open access COVID-19 datasets, which can be used for teaching, training and/or research (depending on their license and available format). We will continually update this page with new databases and registries.

The table below provides a summary of curated COVID-19 imaging datasets. More information on each dataset, including the license, is available by clicking on the name of the dataset or scrolling down.

Last update: March 11, 2021

AIforCOVID imaging archive

Organisation/Curator: CDI Centro Diagnostico Italiano and Bracco Imaging (Milan)

Description: The AIforCOVID imaging archive hosts a large archive of medical images of Italian COVID-19 patients as well as the related clinical data.

Cases: 983 patients

Format: DICOM from chest X-rays

Metadata: Clinical data

Country: Italy

Link: https://aiforcovid.radiomica.it/

License: Data are freely available to browse, download and use for commercial, scientific and educational purposes as outlined in the Creative Commons Attribution 3.0 Unported License.

MosMedData: Chest CT Scans with COVID-19 Related Findings

Organisation/Curator: Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies of the Moscow Health Care Department (MosMed)

Description: This dataset contains axial CT images from patients with COVID-19. Most cases include multiple slices.

Cases: 1110 patients

Format: NIfTI (During the DICOM to NIfTI formatting process only every 10th image was preserved)

Metadata: Clinical data

Country: Russia

Link: https://mosmed.ai/datasets/covid19_1110

License: Creative Commons Attribution Non-Commercial No Derivatives 4.0 (CC BY-NC-SD 4.0)

Italian Society of Radiology COVID-19 database

Organisation/Curator: Italian Society of Radiology (SIRM)

Description: This dataset contains axial CT images from patients with COVID-19. Most cases include multiple slices.

Cases: 68 patients

Format: JPG

Metadata: Clinical data, including PCR status

Country: Italy

Link: https://www.sirm.org/en/category/articles/covid-19-database/

License: authorised for non-commercial research

Radiopaedia COVID-19 database

Organisation/Curator: Radiopaedia.org, compiled by a global team of radiologists and other health professionals

Description: This dataset contains axial volumetric chest CTs.

Cases: 101 patients

Format: JPG

Metadata: Clinical data, including PCR status for some

Country: Global

Link: https://radiopaedia.org/articles/covid-19-3

License: All content is provided under a modified creative commons license. Further details of these modifications are outlined in the Terms of Use.

COVID-19 British Society of Thoracic Imaging imaging database

Organisation/curator: British Society or Thoracic Imaging (BSTI)

Description: This dataset is an imaging database of known UK patients

Cases: 59 patients

Format: online viewer

Metadata: Clinical data, including PCR status

Country: UK

Link: https://bit.ly/BSTICovid19_Teaching_Library

License: free to view and use for education

University of California San Diego COVID-CT database

Organisation/curator: Jinyu Zhao, Yichen Zhang, Xuehai He, Pengtao Xie (all University of California San Diego, US)

Description: This dataset is a collection of open source CT images of COVID-19 cases from preprint articles

Cases: 349 CT images from 216 patients

Format: PNG

Metadata: Medical history, including PCR status, for some cases

Country: Global

Link: https://github.com/UCSD-AI4H/COVID-CT

License: various, depending on source of CT images. Free to use for non-commercial research.

Coronacases.org

Organisation/curator: RAIOSS and Livon Saúde (Brazil), Rodrigo Caruso Chate (Hospital Israelita Albert Einstein, BR)

Description: This dataset contains reviewed CT images of COVID-19 cases

Cases: 10 patients

Format: online viewer

Metadata: Some clinical data, no PCR status

Country: China

Link: https://coronacases.org/forum/coronacases-org-helping-radiologists-to-help-people-in-more-than-100-countries-1

License: listed as authorised for everyone

Eurorad COVID-19 cases

Organisation/curator: Eurorad, a peer-reviewed educational tool of the European Society of Radiology based on radiological case reports to provide a learning environment for radiologists, radiology residents and students worldwide.

Description: This dataset contains chest X-rays and CT scans of COVID-19 patients.

Cases: 50 patients

Format: JPG/PDF

Metadata: Clinical data, including PCR status

Country: Global

Link: https://www.eurorad.org/advanced-search?search=COVID

License: Creative Commons Attribution Non-Commercial Share-Alike 4.0 (CC BY-NC-SA 4.0)

COVID-19 machine learning dataset

Organisation/curator: Joseph Paul Cohen (Université de Montréal, CA)

Description: This is a dataset compiled from various sources, including Eurorad, Radiopaedia, SIRM and various publications. It includes chest X-ray and chest CT images from patients with COVID-19 that were converted to JPG. It also includes 20 CT volmes from Coronacases.orf and Radiopaedia in the Neuroimaging Informatics Technology Initiative format NIfTI.

Cases: 931 images from 461 patient, 20 CT volumes

Format: JPG, NIfTI

Metadata: Clinical data for most, some including PCR status

Country: Global

Linkhttps://github.com/ieee8023/covid-chestxray-dataset

License: various, depending on source of CT images. Free to use for non-commercial research.

COVID-19 CT segmentation dataset 1

Organisation/curator: MedicalSegmentation.com, Håvard Bjørke Jenssen (University Hospital of Oslo, NO)

Description: This is a dataset of axial CT images from patients with COVID-19 that were converted to the Neuroimaging Informatics Technology Initiative format NIfTI from a subset of the openly accessible JPG images from the Italian Society of Radiology.

Cases: 100 axial CT images from >40 patients

Format: NIftI

Metadata: Clinical data, including PCR status, available at original cases from SIRM.

Country: Italy

Link: http://medicalsegmentation.com/covid19

License: authorised for non-commercial research

COVID-19 CT segmentation dataset 2

Organisation/curator: MedicalSegmentation.com, Håvard Bjørke Jenssen (University Hospital of Oslo, NO)

Description: This dataset contains 9 segmented axial volumetric CTs from a subset of cases on Radiopedia. This dataset includes whole volumes with both positive and negative slices (373 out of the total of 829 slices have been evaluated by a radiologist as positive and segmented).

Cases: More than 800 slices from 9 patients

Format: NIftI

Metadata: Clinical data, including PCR status, available at original cases on Radiopaedia

Country: Global

Link: http://medicalsegmentation.com/covid19

License: All content is provided under a modified creative commons license. Further details of these modifications are outlined in the Terms of Use.

COVID-19 CT segmentation dataset 3

Organisation/curator: Ma Jun (Nanjing University of Science and Technology, CN) et al.

Description: This dataset contains labeled COVID-19 CT scans. Left lung, right lung, and infections are labeled by two radiologists and verified by an experienced radiologist.

Cases: 20 patients

Format: NIftI

Metadata: No additional clinical data

Country: Global

Link: https://zenodo.org/record/3757476

License: Creative Commons Attribution Non-Commercial Share-Alike 2.0 (CC BY-NC-SA 2.0)

Please take note of the licenses for each dataset when using it for research and publications. Instructions for references are available on the respective repository’s website for most datasets.

Do not claim diagnostic performance of a model without a clinical study.