Moscow collects largest CT scan database of COVID-19 patients

May 28, 2020

Moscow has the largest database of chest CT scans of COVID-19 patients. It includes over 1,000 anonymised sets of CT scans. These data are used to develop various services based on artificial intelligence.

“Moscow has been fighting the coronavirus pandemic for four months in a row. During this difficult time, artificial intelligence can help doctors: technology can make it possible for them to more quickly and precisely determine the stage of pneumonia in a CT scan of the lungs. At the same time, anonymised data of patients collected between 10 March and 25 April are used to effectively teach artificial intelligence. The database includes over 1,000 scans, which makes it the largest in the world. We hope that the advanced technology we are actively introducing to the Moscow healthcare system will help to counter the infection more effectively,” said Anastasia Rakova, Deputy Moscow Mayor for Social Development.

About 80,000 scans of patients with suspected viral pneumonia taken at Moscow inpatient CT centres make up a database of over 1,000 anonymised scans with special markings that can be used to teach artificial intelligence. These data are available for developers. The scans in the database of the Research and Practical Clinical Centre for Diagnostics and Telemedicine Technology were collected in Moscow between 10 March and 25 April 2020 using the unified radiological information service.

Photo: Press Service of Moscow Mayor and the Moscow Government. Denis Grishkin

“Artificial intelligence is actively used to diagnose the coronavirus infection. The results of all the scans are automatically sent to the unified radiological system. The artificial intelligence processes the data, and the X-ray specialist receives an image with markings and a description as part of the experiment. When preparing a data set, it is important to mark all the data using the localisations and the pattern of the pathology, zones the AI algorithms can use to detect the pathology. During a difficult epidemiologic situation, this helps doctors to notice even the smallest details on the scan,” said Sergei Morozov, chief X-ray and instrumental diagnostics specialist at the Moscow Healthcare Department and director of the Research and Practical Clinical Centre for Diagnostics and Telemedicine Technology.

The data collected by experts of the clinical centre is unique. For example, the database of California’s University of San Diego has 349 2D CT scans of 216 patients, while the data in collected in Moscow include 3D scans. The RAIOSS & Livon Saúde database from Brazil contains 10 CT scans to date, and the Italian Society of Medicine and Interventional Radiology has over 70. The collection of novel coronavirus cases at the Radiological Society of North America is scattered and only suitable for familiarisation, while that of the British Society of Thoracic Imaging has no more than 100 scans.

Photo: Evgeny Samarin

The Moscow database differs from foreign ones in another important way: all the scans are marked according to the classification, which reflects pathological changes in the lung tissue specific to COVID-19. The classification used for the markings was published in the methodical recommendations on X-ray diagnostics of the coronavirus.

To date, Moscow has 48 inpatient CT centres to diagnose patients with acute respiratory virus infection, community-acquired pneumonia and suspected novel coronavirus infection. They are open at city clinics with CT rooms. In addition to CT scans, these centres carry out other tests for patients with suspected COVID-19, such as general blood tests, ECG and swab tests for the novel coronavirus infection. This approach makes it possible to diagnose pneumonia and determine its type.


Latest Events


March 7, 2020

Mariam Merabova and Armen Merabov's project MIRAIF

January 2, 2020
If you continue to use our website, you are agreeing to accept the use of cookies on your device. Cookie files ensure the website’s efficiency and help us provide you with the most interesting and relevant information. Read more about cookie files.
Accept ccokies