London-based researchers have developed a new tool for diagnosing COVID-19 using artificial intelligence (AI).
Scientists at Lawson Health Research Institute developed and trained an artificial neural network to recognize patterns made by the virus in ultrasound lung scans. While essentially a fingerprint of COVID-19, these unique patterns are invisible to the human eye.
"The AI tool that we developed can detect patterns that humans cannot, said Dr. Robert Arntfield, Lawson researcher and medical director of the London Health Sciences Centre (LHSC) critical care trauma centre. "There are details, however, that distinguish COVID-19 at the pixel level that cannot be perceived by the human eye."
Researchers trained the network using more than 100,000 ultrasound lung scans from patients with confirmed COVID-19 infections at the LHSC. Those scans were compared using images from patients with other types of lung diseases and infections.
"The neural network was able to identify the unique characteristics among different scans, and exceed human-level diagnostic specificity," said Arntfield. "While trained physicians could – as expected - not distinguish between different causes of lung disease, the AI had nearly perfect accuracy in making the diagnosis. It’s almost like the AI sees a QR code that we cannot see, unique to the disease.”
The study, published in the BMJ Open, raises the possibility of using AI for other diagnostic purposes, said Arntfield. He has been granted approval for another study examining conditions, other than COVID-19, where AI diagnostic technology and lung ultrasound could be used.