Abstract Introduction COVID-19 pneumonia manifests with a wide range of clinical symptoms, from minor flu-like signs to multi-organ failure. Chest computed tomography (CT) is the most effective imaging method for assessing the extent of the pulmonary lesions and correlates with disease severity. Increased workloads during the COVID-19 pandemic led to the development of various artificial intelligence tools to enable quicker diagnoses and quantitative evaluations of the lesions. Aim of the study This study aims to analyse the correlation between lung lesions identified on CT scans and the biological inflammatory markers assessed, to establish the survival rate among patients. Methods This retrospective study included 120 patients diagnosed with moderate to severe COVID-19 pneumonia who were admitted to the intensive care unit and the internal medicine department between September 2020 and October 2021. Each patient underwent a chest CT scan, which was subsequently analysed by two radiologists and an AI post-processing software. On the same day, blood was collected from the patients to determine inflammatory markers. The markers analysed in this study include the neutrophil-lymphocyte ratio (NLR), monocyte-lymphocyte ratio, platelet-lymphocyte ratio, systemic immune-inflammatory index, systemic inflammation response index, systemic inflammation index, and serum interleukin-6 value. Results There were strong and very strong correlations between the derived inflammatory markers, interleukin-6, and the CT severity scores obtained by the AI algorithm (r=0.851, p<0.001 in the case of NLR). Higher values of the inflammatory markers and high lung opacity scores correlated with a decreased survival rate. Crazy paving was also associated with an increased risk of mortality (OR=2.89, p=0.006). Conclusions AI-based chest CT analysis plays a crucial role in assessing patients with COVID-19 pneumonia. When combined with inflammatory markers, it provides a reliable and objective method for evaluating COVID-19 pneumonia, enhancing the accuracy of diagnosis.
Hălmaciu et al. (Tue,) studied this question.