While we are largely out of the worst parts of the Pandemic, some of the lasting effects of COVID-19 can still be felt, especially in the lungs of those who endured the virus. Studies have found that lasting harm can be left from having the disease and scientists have had a hard time visualizing exactly what is going on to cause the damage. Conventional chest scans cannot reliably detect signs of lung scarring and other pulmonary abnormalities, making it difficult to track the health and recovery of patients.
A research team from King Abdullah University of Science and Technology has recently developed an AI program to change this. Their computer-aided diagnostic tool could be the turning point in finding treatments to diminish the long-term effects of COVID.
Zooming in on the lungs
The method – known as Deep-Lung Parenchyma-Enhancing (DLPE) – layers AI algorithms with standard chest imaging data to reveal previously hidden details in lung dysfunction. Through DLPE augmentation, “radiologists can discover and analyze novel sub-visual lung lesions,” says computer scientist and computational biologist Xin Gao who worked on the project. “Analysis of these lesions could then help explain patients’ respiratory symptoms.”
DLPE works by excluding tissues that are not damaged by COVID-19 from the picture, removing airways and blood vessels to enhance what is left behind. The team validated their technique by using chest scans of thousands of people hospitalized with COVID-19 in China, confirming the tool’s ability to reveal signs of pulmonary fibrosis in those with long COVID.
A new age of lung treatment?
This new technology could allow for better disease management and treatment spanning further than COVID. In the paper, the researchers showed their algorithm could be used as a broad diagnosis tool for various lung problems including pneumonia, tuberculosis, and lung cancer, empowering radiologists to “see the unseen,” as Goa puts it.
Source study: Nature Machine Intelligence – An interpretable deep learning workflow for discovering subvisual abnormalities in CT scans of COVID-19 inpatients and survivors