Scientists at the Belgorod State National Research University (NRU BelSU) have trained a neural network to detect the presence of coronavirus infection in patients using X-ray images. It is reported by TASS with reference to the press service of the university.
To speed up the process of processing and analyzing X-ray images of patients, Russian researchers have developed and trained an ultra-precise neural network, which, by automatically classifying images, is able to quickly detect signs of COVID-19. For her training, the so-called transfer approach was used, which consists of three stages. It involves training a neural network on one dataset and then adapting it to another dataset.
Tests of the neural network have shown its high efficiency. As the assistant professor of the department of mathematical and software information systems of the university Vladimir Mikhelev said, the accuracy of the method reaches 94.17 percent.
Earlier, American doctors named the reason for re-infection with coronavirus. They monitored the health of about 2.5 thousand US Marines from May to November 2020. The study showed that the reason for re-infection may be in weaker immunity.
#Russia #taught #neural #network #detect #coronavirus #Xrays