Abstract
Introduction:Artificial Intelligence (AI) based algorithms demonstrated a higher incidence of COVID-19 pneumonia in the lower lobes, however these algorithms focused on the cluster analysis of pneumonia volumes which may overlooked the pixel-by-pixel CT attenuation changes over the whole lungs particularly the upper lobes since the COVID-19 pneumonia clusters widely developed at peripheral regions at the lower lobes.
Objectives:To develop a technique for quantitative measurement of pneumonia related changes in CT attenuation values over the entire lung in pixel-by-pixel bases rather than only clustered focal pneumonia volumes.
Methods:Total n=100 patients confirmed COVID-19 with RT-PCR (n=50) and age matched healthy cohort (n=50) were analyzed using the proposed technique in this study and a commercially available AI software for comparison. The pixel-by-pixel CT attenuation volumes were calculated by excluding pulmonary airways, vessels, and fissure as shown in Fig. 1.
Results:The %pneumonia from the upper lung lobes in COVID-19 cohort was 6.0±5.2% and %Lung-content from the healthy cohort was 1.3±0.7% and statistically different (p<0.01) which showed a sensitivity 3.9 times higher (p<0.01) than a commercially available AI algorithm.
Conclusion:Using the proposed novel technique, %pneumonia could be calculated not only in the clusters but also upper lung lobes with an improved sensitivity by a factor of 3.9 compared to AI based analysis.
Footnotes
Cite this article as: European Respiratory Journal 2021; 58: Suppl. 65, PA3241.
This abstract was presented at the 2021 ERS International Congress, in session “Prediction of exacerbations in patients with COPD”.
This is an ERS International Congress abstract. No full-text version is available. Further material to accompany this abstract may be available at www.ers-education.org (ERS member access only).
- Copyright ©the authors 2021
