Currently, advanced computed tomography (CT) chest scanning is the only way to investigate persistent respiratory issues following COVID-19 infection. Conventional X-rays can’t provide the needed detail to determine post-COVID lung changes. Yet, CT scans are expensive and often less available than simple X-rays, leaving many Long COVID patients in the dark as to their lung health. A solution to this diagnostic dilemma may be on the horizon, however, thanks to a new scientific advance in lung disease modeling.*

What the researchers did

Scientists from the University of Iowa have developed a deep-learning computer model called contrastive learning. Using diagnostic data from three-dimensional CT images of Long COVID lungs, this method detects and shows lung changes as two-dimensional images. A second method, transfer learning, would then send CT diagnostic information of lung abnormalities directly to a chest X-ray, allowing a simple X-ray to detect lung disease as would an advanced CT scan. 

The researchers tested the model’s accuracy on small airways disease, an early form of lung dysfunction in Long COVID, in a study published in the journal Frontiers in Physiology. They found that:

• 140 post-COVID subjects were analyzed against 105 healthy controls, and
• Features of lung disease in Long COVID patients were detected from CT scans during inhalation and exhalation.

What they found

The model showed 90% accuracy in differentiating Long COVID lung changes from healthy lungs. It was also able to distinguish lung dysfunction severity, separating those with small airways disease from more advanced lung issues like emphysema.

Two distinct clusters of Long COVID patients, or subtypes, were found.

• One had more air remaining after exhalation (air trapping) from small airways disease.
• Another had less lung capacity, and had features consistent with scarring (fibrosis).

Small airways disease affects an intricate web of over 10,000 lung tubules, at the point where oxygenated air enters the bloodstream. Many of these tiny airways are constricted in this condition, reducing oxygen transfer and making breathing difficult.

What it means

Though their study was limited by its small sample size, the researchers say their contrastive learning model was successful in identifying reduced post-COVID lung function, and uncovering two subtypes of Long COVID patients. They say a bigger sample could be useful in uncovering more lung function subtypes.

Their next step is feeding the CT scan information into a model that will translate the lung disease features into two-dimensional images. Said lead scientist Ching-Long Lin, Ph.D., “Clinicians would be able to use chest X-rays to detect these outcomes. That’s the bigger perspective. Chest X-rays are accessible, while CT scans are more expensive and not as accessible. Our model can be further improved, and I believe there is potential for it to be used at all clinics without having to buy expensive imaging equipment, such as CT scanners.”

*University of Iowa. (2022, December 10). Long-COVID’s Effects Can Now Be Detected Using Simple Chest X-Rays. SciTechDaily. https://scitechdaily.com/long-covids-effects-can-now-be-detected-using-simple-chest-x-rays