Artificial intelligence reads CT images to diagnose coronavirus in seconds

A China-based technology company has developed a smart image-reading system that can help radiologists reach a coronavirus diagnosis in seconds, rather than minutes.

Ping An just launched the new artificial intelligence solution on Feb. 19 and its already been used on scans for 5,000 patients, according to a Friday announcement. The company claims its AI tool can pinpoint the disease from computed tomography images in about 15 seconds with 90% accuracy, compared to 15 minutes for a physician.

"Comparing multiple images is a time-consuming task and it cannot be accurately completed manually,” Xiao Jing, PhD, chief scientist at Ping An, said in a statement, adding that the platform can “effectively improve the diagnostic accuracy and the doctor's image-reading efficiency."

Chest CT has been identified as the primary diagnostic tool for this new virus. But in China, where the outbreak originated, there’s been a scarcity of docs to read scans, particularly in areas where the epidemic has hit hardest.

Ping An said it hoped to close this gap by training its smart image-reading system on clinical data from COVID-19 patients. Along with aiding in diagnosis, the “AI analysis engine,” measures changes in lesions, tracks development of the disease and evaluates treatment. Radiology providers can also access the image-reading tool through the cloud, the company noted, and it’s easily transplanted to CT equipment with help from the manufacturer.  

Ping An is based in China’s southern city of Shenzhen, and is one of the largest payers in the world. In addition to insurance, it offers asset management, banking and technology solutions in healthcare and other fields, according to Bloomberg.

More than 84,000 cases of coronavirus have now been confirmed across the globe, with the death toll quickly approach 3,000. Radiologists at Mount Sinai in New York were the first in the U.S. to analyze CT scans from COVID-19 patients, revealing their findings earlier this month in Radiology.

Researchers have released numerous studies since, exploring infection control measures, the imaging physicians’ role in controlling the outbreak beyond making the initial diagnosis, and steps interventional radiologists can take to prepare.