Estimation of Infected Area and its Severity in Chest X-ray Images of Tuberculosis Patients

Authors: Wenfa Ng*

DOI: 10.23880/aabsc-16000228

Volume 7, Issue 2

Abstract

Despite much research, tuberculosis (TB) remains a significant health burden in both developing and advanced countries. Diagnosis of tuberculosis still relies on chest X-ray images which are interpreted qualitatively by radiologist. Hence, much pathological conditions useful for formulating a treatment plan elude detection. This work aims to develop quantitative methods for determining the extent of tuberculosis infection, and quantify the relative proportion of moderate and severe infections in different parts of the lung. Using MATLAB software developed for this study, quantitative analysis of 15 TB infected and 6 latent TB cases reveals higher light intensity in TB infected and latent TB cases relative to normal lung. More importantly, analysis of infected area for pixel intensity > 35 indicates TB infection spread throughout the lung and is considered a systemic disease. In cases of pixel intensity > 90 (moderate to severe), and pixel intensity > 150 (severe) infected area, both TB infected and latent TB presents with these infections to a significant degree, and less burden for latent TB cases. Overall, the top right lung persistently presents with less moderate and severe TB infections, and is the region that can be resuscitated in severe TB pneumonia. Additionally, this work also reveals latent TB as an aggressive lung disease likely kept in check by a relatively strong immune system that prevented whole body physiological collapse.

Keywords: Tuberculosis; Quantitative Image Analysis; Pixel Intensity Analysis; Moderate and Severe Disease; Infected Area

View PDF

View Appendix