COVID‑Net: AI‑assisted diagnosis and prognosis of COVID‑19 infection

A critical step in the fight against COVID‑19 is the effective diagnosis and prognosis of infected patients. This project leverages x‑ray, CT scan and ultrasound images, which are readily available in hospital settings, to provide new screening options for rapid triaging in parallel to viral testing. In the case of ultrasound, the technology is non‑invasive, mobile, and increasingly accessible to remote and low‑resource communities.

Researchers from the University of Waterloo and the National Research Council of Canada (NRC) are using artificial intelligence (AI) to aid in the patient chest imaging screening process during the COVID‑19 clinical workflow. AI can play a key role in looking at medical images and helping to determine the incidence and severity of COVID‑19 infection.

The outcomes of this project include open access AI models and the largest, fully curated databases of systematically processed images. These tools could help hospitals, medical device and software firms develop new AI screening solutions for the current and future pandemics, and for other lung infections.

Collaborators

• Research: University of Waterloo, the NRC's Digital Technologies Research Centre
• Data: National Institutes of Health, hospitals, publicly available datasets
• Computation power: DarwinAI, NVIDIA, Hewlett Packard Enterprise, Microsoft
• End products and users: DarwinAI, Red Hat, Boston Children's Hospital
• Funding: DarwinAI, the NRC's Pandemic Response Challenge Program

Objectives

• Collect and curate data (images, video) of x‑ray, CT scan and ultrasound images of COVID‑19‑positive and COVID‑19‑negative patients
• Develop models for risk stratification and diagnosis using few‑shot learning and meta‑learning
• Develop models for lung segmentation and cropping
• Produce new AI tools to assist with COVID‑19 diagnostics and prognostics

Deliverables

• Open access x‑ray dataset and models for COVID‑19 detection
• Open access CT scan dataset and models for COVID‑19 detection
• Open access ultrasound dataset and models for COVID‑19 detection

Project team

• Alex Wong, Canada Research Chair in Artificial Intelligence and Medical Imaging, University of Waterloo
• Ashkan Ebadi, Senior Research Officer, Data Analytics Centre
• Pengcheng Xi, Senior Research Officer, Computer Vision and Graphics

Publications

• Understanding the temporal evolution of COVID‑19 research through machine learning and natural language processing
• COVID‑19 detection from chest x‑ray images using deep convolutional neural networks with weights imprinting approach