The role of artificial intelligence technology analysis of HRCT images in predicting the severity of COVID-19 pneumonia.

ABSTRACT

INTRODUCTION: The purpose of the study was to analyze the role of automatic assessment of COVID-19 pneumonia severity in high resolution computed tomography (HRCT) images by artificial intelligence (AI) technology. PATIENTS AND METHODS: We retrospectively studied the medical records of consecutive patients admitted to the Krakow University Hospital due to COVID-19. Of the 1,729 patients, 804 had HRCT with automatically analyzed radiological parameters absolute inflammation volume (AIV), absolute ground glass volume (AGV), absolute consolidation volume (ACV), percentage inflammation volume (PIV), percentage ground glass volume (PGV), percentage consolidation volume (PCV) and severity of pneumonia classified as none, mild, moderate, or critical. RESULTS: The automatically assessed radiological parameters correlated with the clinical parameters that reflected the severity of pneumonia (p < 0.05). Patients with critical pneumonia, compared to mild or moderate, were more frequently men, had significantly lower oxygen saturation, higher respiratory rate, higher levels of inflammatory markers, more common need for mechanical ventilation, and admission to the intensive care unit (ICU); moreover, they were more likely to die during hospitalization. Notably, as determined by the receiving operating characteristic curve, radiological parameters above or equal the cut-off points were independently associated with in-hospital mortality (ACV odds ratio (OR) 4.08, 95% confidence limits (CI) 2.62 - 6.35; PCV OR 4.05, CI 2.60 - 6.30). CONCLUSIONS: Using AI to analyze HRCT images is a simple and valuable approach to predict the severity of COVID-19 pneumonia.