Framework for Effective Image Processing to Enhance Tuberculosis Diagnosis

Abstract

Tuberculosis (TB) is one of the disease causing microorganisms which is infectious with a high morbidity and mortality around the world. According to the latest World Health Organization (WHO) report, there were an estimated 10.4 million new TB cases in 2015 and more than 1.8 million deaths were attributed to the disease the same year. A person with TB causing bacteria need to know the presence of the bacteria at its latent (passive) stage so that she/he can take medication only for a couple of weeks to a couple of months (daily, bi-weekly or weekly depending on the treatment regimen of choice) to be free of TB. A person who is diagnosed with an active form of TB has to take medication for a minimum of 6 to 9 months. Negligence to take this medication properly will make the TB causing bacteria drug resistant. Diagnosing TB causing bacteria in its different stages is an effective mechanism which can enhance the performance of present detection schemes available in different clinics. In that regard, this paper presents a methodology for use in enhancing TB diagnosis specificity based on image processing of lung images acquired using normal x-ray as well as those from a sputum smear microscopy. A general framework is designed preceded by a step to extract useful imaging features for use in robust characterization of latent, active, drug resistant and TB free samples.