Methods of Interpretation of CT Images with COVID-19 for the Formation of Feature Atlas and Assessment of Pathological Changes in the Lungs

Zotin, A.; Kents, A.; Simonov, K.; Hamad, Y.

The paper is devoted to the methods of interpretation and analysis of dynamic changes associated with COVID-19 in CT images of the lungs. An attempt was made to identify possible regularities of the CT pattern and diagnose the possible development of fibrosis in the early stages. To improve the accuracy of diagnosis and prognosis of the formation and development of fibrosis, we propose the creation of the Feature Atlas of CT images with a specific X-ray state. An experimental study within the framework of the formed dataset, divided into 4 groups according to the severity of changes, was carried out. The preliminary results of processing and texture (geometric) analysis of CT images were obtained. The analysis of a series of CT images includes key steps such as preprocessing, segmentation of lung regions and color coding, as well as calculation cumulative assessment of features to highlight areas with probable pathology, combined assessment of features and the formation of the Feature Atlas. We generated the preliminary Feature Atlas for automation and more accurate analysis of the CT images set. As part of the study on the selected groups of patients, the areas with the probabilities of pathologies associated with COVID-19 development were identified. The study shows the dynamics of residual reticular changes. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.