In a new study published in the journal Frontiers in Physiology, it was reported that researchers from the University of Iowa have developed a "contrastive learning" model that "learns" from composite 2D images constructed from 3D CT images and detects the compromised lung functions in long-term COVID patients. By adding another technique to 3D CT images, called transfer learning, the researchers were able to convey and connect the results of the lung diagnostic images from a CT scan to a chest X-ray. This allowed them to detect abnormalities in lung function.

The lead author of the study stated that the new element in the model acquires the needed information from 3D CT scans, such as lung volume, and transfers that information to a model that will show the same characteristics in 2D images. In the study, the researchers compared the results of the learning model with a control group that had not contracted the virus. They observed that the advanced model was able to separate patients with small airway disease from those with more advanced complications, such as emphysema.

Additionally, based on the findings of the study, they revealed that patients with post-COVID have two types of lung injuries, namely, small airway disease and lung parenchymal fibrosis or inflammation. They noted that these manifestations of the infection were persistent even after recovering from the SARS-CoV-2 infection. 

(Source: https://health.economictimes.indiatimes.com/news/diagnostics/researchers-detect-long-covids-impact-via-simple-chest-x-rays/95286352 )