Avaliação entre a correlação de atributos das áreas alteradas imagens de ressonância magnética de pulmão, bem como a concentração de marcadores inflamatórios no sangue de pacientes com COVID-19

Abstract

SARS-CoV-2, also known as COVID-19, was responsible for the global pandemic that started in 2019, which has repercussions to this day. This project aimed to find a correlation between characteristics extracted from the inflamed regions present in pulmonary magnetic resonance imaging and inflammatory indicators from blood tests of a set of patients diagnosed with COVID-19, through image processing and data mining techniques. Among the methods used to perform this analysis are correlation mining, association rules and classification through decision tree. From the results obtained, it was possible to find a correlation between the study variables, where a high rate of the inflammatory markers urea and creatinine in the blood is associated with one of the characteristics of pulmonary magnetic resonance imaging, which is the number of pixels present in the inflamed regions. However, this correlation alone does not conclusive the possibility of using magnetic resonance imaging as a substitute method for those currently used for the diagnosis of patients with COVID-19. Other analyzes such as patient classification, using correlation, to identify who needs to perform the blood test, in addition, it is possible to create new analyzes improving the techniques used for data extraction and even opting for the extraction of other attributes from the images.