Harnessing the power of machine learning to help clinicians more easily identify lung abnormalities in chest X-rays.
Project Budget*- $2.8M
Partner Co-investment - $0.8M
Supercluster Co-investment - $2.0M
In the fight against COVID-19, chest x-rays have been a critical tool for identifying lung abnormalities in patients, a critical complication from COVID-19. Chest x-rays can show the signs of COVID-19 infection in the lungs in multiple ways, such as partially collapsed lungs, blocked bronchial tubes and liquid collecting in the lungs or their outer lining. The information is key for family doctors, emergency doctors and nurses trying to determine if a patient could be infected.
The challenges faced by frontline teams trying to detect and treat COVID-19 include the time it takes to receive a formal report from a radiologist, along with the accurate identification of abnormalities in the lung. Often the treatment teams rely on their own interpretation of an x-ray to manage a patient while waiting for a radiologist’s report, if they have access to one. These interpretations drive clinical decisions, including the decision to admit or discharge a patient. A late or incorrect diagnosis could result in an infectious patient being sent home and a rapid deterioration of their health.
XrAI is an AI-driven chest x-ray tool developed by 1QBit in close partnership with Canadian healthcare organizations and physicians. XrAI has been approved by Health Canada as a Class III medical device to support Canadian health providers in the fight against COVID-19. This software is ready to be deployed across Canada through the project team led by 1QBit in partnership with Saskatchewan Health, British Columbia’s Fraser Health, Vancouver Coastal Health and First Nations Health authorities, Ontario’s Trillium Health Partners, and Microsoft.
XrAI integrates seamlessly into existing clinical information systems. There are no interruptions to workflow and little to no training required upon installation. XrAI’s rapidly delivers AI-driven findings directly to clinicians’ standard x-ray viewer.
The technology already has proven successful. It has been tested on publicly available data of COVID-19 pneumonia and it correctly identified 100 per cent of the cases presenting abnormalities in the lungs. Furthermore, in a randomized control trial, the solution improved the accuracy of identifying other lung abnormalities across a variety of physician groups.
Chest x-rays are already performed in 35 per cent of visits to the ER – and COVID-19 is driving higher usage. By speeding up the analysis of the x-rays, XrAI will help reduce radiology bottlenecks and will be especially valuable for rural and remote ERs and doctors.