Objective quantification of chest radiography exams for pulmonary fibrosis

Abstract

Tuberculosis, caused by Mycobacterium tubercu-losis, is an infectious disease which remains a global health problem. The chest radiography is the common method employed to assess the tuberculosis evolution. Despite the tuberculosis diagnosis be normally realized by chest radiography analyses, the objective quantification of pulmonary diseases has been devel-oped with high resolution computed tomography scans. This quantification is important to assess evolution, treatment and comparison of different tuberculosis treatments. The aim of this work is to develop a methodology for pulmonary fibrosis quanti-fication caused by tuberculosis using chest radiographs. An algorithm was developed in Matlab® for computational pro-cessing of exams, which creates a lungs’ 3D representation with fibrosis regions. The developed method was performed with a group of 14 patients diagnosed with pulmonary Tuberculosis. For the same patients, the data were compared with quantifica-tion method of high resolution computed tomography scans, previously described in the literature. Statistical analysis was performed by linear regression and Bland-Altman plot. The results showed an average variation of 11.32% between the two quantification methods. The results suggest the effectiveness and applicability of the developed method, providing better risk-benefit to the patient and cost-benefit ratio for the institution.