Rapid testing and the use of PCR tests may be free or easily accessible to Canadians to scan for COVID-19, but that is not the case in other countries.
UBC Okanagan researchers say rapid tests can be limited and very expensive in developing countries, which is why they are using artificial intelligence to find another method to test for the virus.
Dr. Mohamed Shehata, an associate professor of computer science at UBCO, along with his team, have developed ‘CORONA-Net’, a software that can quickly and easily detect COVID-19 using chest X-ray images.
“There are X-ray machines already in these countries because that’s part of the normal medicine practice. X-rays are available and they are very cheap. So the X-ray is considered an alternative method that is reliable and accurate, sometimes even more than the PCR test”, says Dr. Shehata.
Although X-rays are already available in other countries to detect COVID-19, it takes time for the image to be evaluated by a physician and for the infection to get detected — which can result in further spread of the virus. However, by using ‘CORONA-Net’, the artificial intelligence system can flag suspicious cases to be fast-tracked.
Get weekly health news
“The software that we developed, we wanted to have 100 per cent detection of COVID-19. The overall accuracy was about 95 per cent, but our main focus was we did not want to miss anything even if it turns out later to not be COVID-19. The big thing with this software is each time it gets the images, it can learn and increase its accuracy,” explains Dr. Shehata.
Dr. Shehata adds that the PCR method is unable to detect if someone previously had the virus, something the X-ray imaging is capable of doing.
“According to recommendations by the World Health Organization (WHO), X-ray imaging is an effective method for detecting if people have been affected or have recovered from the virus.”
Dr. Shehata is in talks with other countries in order to have ‘CORONA-Net’ in use to make rapid testing more widely available.
The research paper was published in the Journal of Imaging.
Comments