In the fight against COVID-19, the most crucial weapons will be quick screening tools. A neural network powered technique called COVID-Net, introduced earlier this week could be the perfect tool researchers have been seeking all this time. COVID-Net will be able to aid scientists to develop AI tools that can perform quick COVID-19 diagnosis based on images of chest X-rays. 

The recent struggle faced by the medical community to provide quick COVID-19 screening based on legacy methods have led to the rise of alternative screening methods such as radiography examination. In radiography examination, chest X-rays and CT scans are analysed for visual indicators associated with SARS-CoV-2 viral infection. It is based on the earlier findings that patients infected with COVID-19 show specific lung abnormalities in their chest X-rays. This has led many researchers to suggest that radiography examination should be used as a primary tool for COVID-19 screening in epidemic areas.

Coming to COVID-Net, it is a deep convolutional neural network design tailored for the detection of COVID-19 cases from chest radiography images that are open-source and available to the general public. It is developed by Linda Wang and Alexander Wong from the University of Waterloo and DarwinAI in Canada. The findings were presented in a paper titled “COVID-Net: A Tailored Deep Convolutional Neural Network Design for Detection of COVID-19 Cases from Chest Radiography Images” uploaded on March 22 to arXiv -the open-access repository of pre-print research papers.

A dataset called COVIDx, comprised of 5941 posteroanterior chest radiography images across 2839 patient cases, with various lung conditions, including bacterial infections, non-COVID viral infections, and COVID-19, was used to train the COVID-Net. The dataset is also made publicly available for other researchers to use. 

On the other hand, the researcher does raise caution, suggesting that it is “by no means, a production-ready solution” and ask other researchers’ help to turn it into one.

“COVID-Net has yet to prove itself, but it follows in the footsteps of a previous success story.” writes Will Douglas Heaven on Technology Review. 

“Many of the big advances in computer vision in the last ten years are thanks to the public release of ImageNet, a large data set of millions of everyday images, and AlexNet, a convolutional neural network that was trained on it. Researchers have been building on both ever since”, adds Will Douglas Heaven.