Artificial intelligence (AI) has been taught to diagnose for coronavirus infection by examining CT scans of patients’ lungs, according to a report from US-based magazine The Scientist.
In China, an AI was fed 532,000 CT scans from 3,777 patients to look for coronavirus traits in lungs, the magazine reported. Using this AI model, hospitals in China were able to correctly diagnose patients with coronavirus 85 percent of the time.
Coronavirus testing has been identified as critical by governments and health authorities around the world, as it allows for health authorities to implement contact tracing with a high degree of accuracy, limiting the spread of the virus; it also helps governments to know when it is the right time to ease lockdown measures.
Many patients suffering from the coronavirus develop pneumonia, which can lead to death. This pneumonia is also different from more common types of bacterial pneumonia, The Scientist reported, which allowed researchers to train AI to find the virus.
“Most striking are cloudy lesion patterns that resemble shards of glass or reticular lines within the opaque lesions that look like irregular paving tiles, which occur around the peripheries of both lungs. Lesions from bacterial pneumonia are usually concentrated in one lung and may not resemble shards of glass,” the magazine said.
Noting the difference between pneumonias will be particularly pertinent as the coronavirus crisis recedes and pneumonia caused by the virus is no longer the leading type of pneumonia among patients, The Scientist reported.
Accurately diagnosing coronavirus pneumonia will allow health professionals to tailor treatment to the virus.
The AI model is already at use in 10 large hospitals in China, and several others in the US, India, Iraq, and Ecuador, The Scientist said.