VANCOUVER -- Combine thousands of chest images of COVID-19 patients from around the world with artificial intelligence, and you’ll get what scientists believe will be latest weapon in the fight against the coronavirus.
Vancouver doctors and researchers are collecting the samples – CT scans and X-rays from as far away as China, Italy and the Middle East – and feeding them into a computer.
“CT is a powerful problem solving tool,” said Dr. Savvas Nicolaou, director of emergency and trauma radiology at Vancouver General Hospital.
The goal? Instant diagnosis of things like the severity of the illness and what treatment would work best.
“They can actually assist physicians in (determining) which patients can be safely discharged back home, or which patients might need to be admitted, require ICU, or require ventilation,” said Nicolaou, who is also working on the project with researchers at UBC.
“The healthy lung looks very dark, because the primary component of a lung is air, and you can’t see air” he said. On the other hand, a COVID-19 lung will have lots of white spots and look like it’s filled with snow.
The team envisions a sick patient getting CT scan, which can be done quickly, and then a machine would provide instant results.
So far, doctors can’t predict COVID-19’s severity in patients of different ethnicities and genders, because most algorithms have only been studied in one ethnic population. Researchers are confident this new tool will help fix that.
“We’re hoping we can develop these intelligent, automated algorithms for the future,” Nicolaou said.
The Vancouver team is also working on “virtual twins,” which would allow doctors to see how a patient would react to treatment before actually giving it to them, by comparing them to a model of themselves created using the compiled data from other, similar patients.
“So we don’t actually cause any harm to yourself, we will see what happens in the virtual twin, and that’s the ultimate goal,” said Nicolaou.
