| Abstract |
Introduction: An early diagnosis of Coronavirus Disease (COVID19) is of utmost importance, so that patients can be isolated and treated in time, eventually preventing spread of the disease, improving the prognosis and reducing the mortality. High Resolution Computed Tomography (HRCT) chest imaging and Artificial Intelligence (AI) driven analysis of HRCT chest images can play a vital role in management of COVID-19 patients. Aim: To explore the various HRCT chest findings in different phases of COVID-19 pneumonia and to assess the potential role of AI in quantitative assessment of lung parenchymal involvement in COVID-19 pneumonia. Materials and Methods: The present retrospective observational study which was conducted between 1st May 2020 to 13th August 2020. Reverse Transcription-Polymerase Chain Reaction (RT-PCR) positive 2169 COVID-19 patients who underwent HRCT chest were included in the study. Presence and distribution of lesions like: Ground Glass Opacity (GGO), consolidation and any specific patterns like septal thickening, reverse halo, sign, etc., were noted in the HRCT images. HRCT chest findings in different phases of disease (Early: <5 days, Intermediate: 6-10 days and Late phase: >10 days) were assessed. CT Severity Score (CTSS) was calculated based on the extent of lung involvement on HRCT, which was then correlated with the clinical severity of the disease. Artificial Intelligence powered “CT Pneumonia analysis” algorithm was used to quantify the extent of involvement of lungs by calculating Percentage of Opacity (PO) and Percentage of High Opacity (PHO) in lungs. Tests of statistical significance, like Chi-square, Analysis of Variance (ANOVA) and Posthoc tests were applied depending on the type of variables, wherever applicable. Results: Radiological findings were seen in HRCT chest of 1438 patients. Typical pattern of COVID-19 pneumonia, i.e., bilateral, peripherally located GGO with or without consolidation was seen in 846 patients. About 294 asymptomatic patients were found to be radiologically positive. HRCT chest in the early phase of disease mostly showed GGO. Features like increased reticulation, predominance of consolidation, presence of fibrous stripes indicated late phase. About 91.3% of cases having CTSS ≤7 were asymptomatic or clinically mild whereas, 81.2% cases having score ≥15 were clinically severe. The mean PO and PHO (30.1±28.0 and 8.4±10.4, respectively) were remarkably higher in clinically severe category. Conclusion: Progression of COVID-19 pneumonia is rapid, so radiologists and clinicians need to get familiarised with the typical CT chest findings, hence patients can be treated on time, eventually improving the prognosis and reducing the mortality. Artificial Intelligence has the potential to be a valuable tool in management of COVID-19 patients.
|
