UC San Diego Uses AI to Rapidly Detect COVID-19 Pneumonia in Lung Scans
UC San Diego Health physicians utilize an artificial intelligence algorithm to quickly analyze chest X-rays for early signs of pneumonia, aiding in the detection and management of COVID-19 patients.
UC San Diego Health radiologists use artificial intelligence to accelerate lung imaging analysis and quickly detect pneumonia in COVID-19 patients. Powered by Amazon Web Services, this technology helps physicians distinguish which patients require intensive hospital care and which ones can safely recover at home. The AI system has already provided unique insights into more than 2,000 medical images.
The algorithm highlights subtle signs of pneumonia by overlaying chest X-rays with color-coded maps that indicate the probability of the condition. It learns from 22,000 notations made by human radiologists to identify inflammation and fluid buildup that might otherwise go unnoticed. In one instance, the AI spots early pneumonia in an asymptomatic emergency department patient, leading to a positive COVID-19 test and immediate medical intervention.
This early detection capability allows doctors to identify high-risk patients before severe symptoms develop or a stethoscope picks up the problem. By catching pneumonia early, the medical team aims to treat those most at risk for severe disease and death more effectively. While the system remains investigational, it already influences the clinical management of patients at UC San Diego Health.