Researchers training artificial intelligence to diagnose COVID-19 –

Original article was published on artificial intelligence

By May 20, 2020 11:08am

Researchers at Lawson Health Research Institute are testing whether artificial intelligence (AI) could be used to help diagnose COVID-19.

Using ultrasound lung scans of patients already confirmed to have the virus at the London Health Sciences Centre, scientists are trying to train an AI system to recognize unique patterns in the imaging.

“Machines are able to find patterns that humans cannot see or even imagine. Lung ultrasound scans of patients with COVID-19 pneumonia produce a highly abnormal imaging pattern. This pattern isn’t unique to COVID, and can be seen in other causes of pneumonia,” Lawson Researcher Dr. Robert Arntfield said in a statement.”It is plausible, however, that there are details that distinguish COVID-19 at the pixel level that cannot be perceived by the human eye.”

If the AI system is able to successfully identify unique characteristics among different scans, the technology could be applied to portable ultrasounds to make bedside diagnoses of the disease, said Arntfield.

Lung ultrasounds have already proven to be highly accurate when diagnosing different types of respiratory infections, such as pneumonia. Ultrasounds are more portable and cheaper than using other diagnostic imaging machines, making them a standard bedside tool in hospitals around the world.

The study into AI’s ability to diagnose COVID-19, is the latest undertaking of the team of London researchers focused on using the technology to solve sophisticated problems.

“This project is a great example of the unique ability we have here in London to be agile: that is, to identify a gap and move quickly towards finding a solution,” said Arntfield, who is also medical director of the critical care trauma centre at LHSC. “I am thrilled that we were able to move through the approval process quickly, and get our ideas working in such a short amount of time.”