Abstract—Infectious diseases are major health related problem in the world. Tuberculosis is included among the infectious, penetrating and prevalent disease caused by mycobacterium tuberculosis. Tuberculosis mostly effects lungs but also occurs in other parts of the body. The screening of tuberculosis is a tiresome task that requires highly trained workers. Screening and diagnosis of tuberculosis has been made possible by the discovery of X-ray radiology. The objective of this study is to summarize different automated tuberculosis detection approaches using chest X-ray and microscopic images. The paper present an automated system designed for screening of tuberculosis using CXR’s. A total of 50 sample images were taken. Algorithm was implemented using MATLAB version 9.2. The interpretation of results was done using PASW statistics version 18. The accuracy of 92% was obtained for chest radiography algorithm that is above 90% so the algorithm can be used for screening of tuberculosis patients. Secondly, the digitize image is an important means for screening, diagnosis and treatment of tuberculosis. The proposed system will result in designing an efficient and accurate system with comparison to a manual system. In future these approaches can be used to other laboratory based techniques of tuberculosis screening e.g. smear microscopy and Drug Susceptibility test.

Index Terms—CAD, CXR, ROC, P-value.

Sana Fatima and Syed Irtiza Ali Shah are with National University of Sciences & Technology, Islamabad, Pakistan, Pakistani. Muhammad Zia Samad is with National TB control program, Islamabad.

Cite:Sana Fatima, Syed Irtiza Ali Shah, Muhammad Zia Samad, "Automated Tuberculosis Detection and Analysis Using CXR’s Images," International Journal of Computer and Electrical Engineering vol. 10, no. 4, pp. 284-290, 2018.