A review on automatic tuberculosis screening using chest radiographs

Tuberculosis (TB) these days is considered as a major health threat in most of the countries of the world. Bacillus, also referred to as Mycobacterium tuberculosis, is the main cause of mortality due to TB. It is reported that mortality rates of patients with tuberculosis are higher when it is not diagnosed at an early stage. At present the approaches used to diagnose TB are based on the incorrect approaches that were developed in the last century. In order to Automate the process and correctly diagnose TB variety of new approaches have been proposed using Computer Aided Diagnostics (CAD) systems. This paper will review various systems that have been proposed for early detection of tuberculosis using chest radiographs, that enable timely treatment. The study focus on segmentation techniques, extraction of features and classification of X-rays as normal or abnormal on the basis of classier that is trained on number of features. Analysis of different approaches based on the performance achieved on benchmark data sets and different parameters is conducted.