DarwinAI, University of Waterloo develop neural network for COVID-19 detection


Kitchener-Waterloo startup DarwinAI, which aims to help developers accelerate deep learning development, has co-developed a neural network for COVID-19 (coronavirus) detection via chest radiography.

“We hope … we can attract clinicians and scientists far and wide to improve upon the technology.”

The tool, called COVID-Net, was created in collaboration with the University of Waterloo’s Vision and Image Processing (VIP) Lab. An important step in the fight against COVID-19 is the effective screening of infected patients, with one of the key screening approaches being radiological imaging using chest radiography.

Linda Wang and Alexander Wong of the VIP Lab said in a post on GitHub that a number of AI systems based on deep learning have been proposed to accurately detect COVID-19 in patients using chest radiography images, but many of those networks have been closed to the public.

DarwinAI’s tool is aimed to accelerate the development of highly accurate and practical deep learning solutions for detecting COVID-19 cases, in a way that is open and accessible to the public. DarwinAI’s dataset includes 5,941 chest radiography images across 2,839 patient cases. The startup is open-sourcing this model so the ecosystem can create a tool to assist health care professionals in combating the pandemic.

RELATED: Here’s how #CDNtech companies are pitching in during COVID-19

“The global crisis brought on by COVID-19 has affected us all,” DarwinAI CEO
Sheldon Fernandez wrote in a company blog post. “Like many businesses, we’ve been grappling with how to best deploy our skills in service of the present crisis.”

COVID-Net was created by Linda Wang and Alexander Wong at the University of Waterloo, as well as DarwinAI, which came out of stealth mode with a $3.9 million CAD in seed round in 2018. To develop the tool, the researchers used public sources to create a dataset of radiography images.

“This software has had promising initial results,” Wong said. “We hope that by making this software open, we can attract clinicians and scientists far and wide to improve upon the technology.”

Several researchers have created tools to help fight the pandemic. Last week, DNAstack launched a new tool for scientific and medical communities to share and discover knowledge about the genetics of COVID-19.

DarwinAI and the VIP Lab are also launching an explainability tool that shows how their AI technology reaches its COVID-19 detection decisions. The organizations are asking researchers or clinicians that would like access to the explainability tool to assist, or have COVID-19 data to share, to contact DarwinAI.

Image source DarwinAI

Isabelle Kirkwood

Isabelle Kirkwood

Isabelle is a Vancouver-based writer with 5+ years of experience in communications and journalism and a lifelong passion for telling stories. For over two years, she has reported on all sides of the Canadian startup ecosystem, from landmark venture deals to public policy, telling the stories of the founders putting Canadian tech on the map.

0 replies on “DarwinAI, University of Waterloo develop neural network for COVID-19 detection”