Approbation of the shearlet transformation method for visualization of pathological changes in the lungs on CT images for diagnosing COVID-19
Radiation diagnostics is a rapidly developing field of medicine which actively includes such concepts as artificial intelligence, computer vision and new methods of medical imaging. Given the urgency of the problem of the appearance of Covid-19 a methodology for processing, analyzing and interpreting CT images is proposed for the effective detection, texture analysis and visualization of pathological changes in the lungs with Covid-19. In the format of advances in AI and computer vision in diagnostics, combined in a new direction – radiomics which is based on the selection of a set of quantitative parameters of the pathology under study with the most accurate values of indicators (markers). Depending on the purpose of the medical research, the extracted features (markers) will differ. An analysis of textural features was carried out based on spectral decomposition methods (wavelet and shеarlet transform of images) with their contrasting with color coding. This approach makes it possible to more accurately assess the quantitative characteristics of the identified changes. As a result of experimental studies a presentation was formed for a medical specialist, followed by a final X-ray diagnostic conclusion. The study was carried out within the framework of the grant «Methods of artificial intelligence and computer vision to improve the accuracy of remote diagnostics of respiratory diseases in the northern group of regions of the Krasnoyarsk Territory» with financial support from the Krasnoyarsk Regional Fund for the Support of Scientific and Scientific and Technical Activities.