ORLANDO, Fla. — Imagine if COVID-19 could be detected virtually.
A reasearcher from the University of Central Florida is part of a breakthrough study doing just that, by using artificial intelligence to spot COVID-19 in a person’s lungs.
For the past decade, Dr. Ulas Bagci has been developing computer processes — or artificial intelligence — to help doctors in the difficult detection of pancreatic and lung cancer.
His latest project is applying that expertise to COVID-19.
A longtime collaborator with the National Institute of Health, Bagci teamed up with NIH researchers to create software to recognize the virus.
To do so, they scanned the lungs of more than a thousand patients in China, Japan and Italy, and then tested their formula on more than a thousand more patients with lung diseases including COVID-19 and cancer.
They found an up to 90% success rate in spotting COVID-19 pneumonia in CT scans. They could even distinguish it from other conditions like the flu and cancer.
“Our idea is to utilize these technologies in the clinic and really go beyond mathematical gymnastics, and just make it really useful for doctors to use and identify those patients,” Bagci said.