Researchers at the Center for diagnostics and telemedicine of the Moscow Department of health, formed the world’s largest database of CT studies of the chest patients with signs COVID-19. The database contains thousands of fully anonymised sets of computer tomograms. The dataset used for developing different services based on artificial intelligence (AI), said Deputy Moscow mayor for social development Anastasia Rakova.
For formation of a base anonymised and marked images on the basis of which it is possible to train artificial intelligence, we have analyzed about 80 thousand KT, made in Moscow outpatient CT centers patients with suspected viral pneumonia. Dataset available for developers of artificial intelligence. CT scan converted into a recognized among professionals machine learning NIfTI format.
Dataset collected by the experts of the Center of diagnostics and telemedicine, unique, stress in the capital daskale. For example, in the database of the University of California in San Diego is 349 two-dimensional CT images of 216 patients, while in dataset, assembled in Moscow, provides three-dimensional CT study. Database RAIOSS & Livon Saúde, formed in Brazil, yet includes 10 computer tomograms, whereas the Italian radiological society, it contains over 70. A set of cases of new coronavirus infection of North American radiological society fragmented and is only suitable for review, and the British society of thoracic radiology it contains more than a hundred studies.
In addition to the number of cases, the Moscow database has another difference from the foreign ones. All CT studies in dataset Center for diagnosis and telemedicine are marked according to the classification reflecting the amount of pathological changes in lung tissues Covid-19 the CT scan of the chest. The classification that formed the basis of markup, has been published in the methodical recommendations for the diagnosis of radiation coronavirus disease Covid-19.
Studies included in the database of the Centre for diagnostics and telemedicine, have been collected in Moscow for the period from 10 March to 25 April 2020 using a Single radiology information service (RICE).