The Celsus team has completed the development of the first version of the system that can analyze lung CT scans and detect signs of coronavirus 2 (SARS-COV-2).
Russian and foreign data sets were used to teach the neural network. Old CT studies with no signs of viral pneumonia, made in 2019 and earlier, are also used as examples without COVID-19.
According to epidemiologists, no sooner is the vaccine developed and clinically tested than we can speak about the victory over COVID-19. Until then, it is important to focus on diagnosis and treatment of this disease.
“The idea to start working towards a solution for coronavirus has been around for quite a long time. But we didn’t want to hype and get to work until it was confirmed that CT scanning can be used to diagnose this disease effectively. Also, we were waiting for enough data to be available. And finally, when the stars aligned, we got to work,” says Evgeny Nikitin, Chief technology officer.
At the moment, image segmentation for labeling affected areas is under development. Improvement of architecture and operating speed of the system is in progress.
In the future, the Celsus will be able to detect not only COVID-19, but other diseases as well. We are now teaching the neural network to detect signs of oncology and different types of pneumonia on tomography.