COVID-19

COVID-19 results in a multicompartmental disease in the lungs, which not only involves lung parenchyma, the portion of the lung involved in gas transfer, but also affects pulmonary vasculature.

The aim of our study is to evaluate regional pulmonary vasculopathy using artificial intelligence.

There is physiologic and imaging data indicating presence of pulmonary angiopathy in COVID-19 that ultimately leads to defects in the lungs. However, this aspect of disease process in COVID-19 is not very well understood and requires further research.

Our retrospective single centre study consists of patients with confirmed COVID-19 on reverse transcription polymerase chain reaction (RT-PCR) who underwent unenhanced CT chest followed by CT pulmonary angiogram (CTPA) during the first wave of the pandemic.

The study is currently at analysis stage, looking at the relationship between visual annotations performed by chest radiologists and algorithm output. AI would quantify the amount of diseased lung versus healthy lung, which will then be correlated with the visual scores performed by radiologist and other clinical parameters.

The results of this study can help facilitate future research to better understand the disease process in COVID-19. This can lead to target therapy to minimise vasculopathy in COVID-19 by means of supressing specific overdriven immune pathways to minimise their adverse impact on pulmonary vasculature. The results of this study could also be applicable to a large group of diffuse lung disease including interstitial lung diseases and diseases with pulmonary vasculopathy such as sickle cell lung disease.

The project is a collaboration between the Clinical Radiology Department at King’s College Hospital and The London Medical Imaging & AI Centre for Value Based Healthcare.