Mycobacterium Tuberculosis Detection using Support Vector Machine Classification Approach

摘要

Tuberculosis (TB) is a transferable malady caused by Mycobacterium Tuberculosis (MTB) and invades when the infected people sneeze, cough and speak without mask. The germs can exist in the air for some hours that consequence persons who breathe in the air may become infected that mainly influence the lungs; it also influenced other limbs of the body such as spine, kidneys and brain. TB malady is very common due to the weakening of the immune system and the possibility of patient’s death enhances with time if left undiagnosed. Conventional TB diagnosis requires much time and money because of microscopic examination of sputum, so automatic recognition is more valuable to prevent serious consequences rather than manually. In this framework, we present an efficient approach for automatic identification of Tuberculosis using some image processing techniques. The role of DIP in medical is more prominent. Usually, most of the methods may deform the authentic information that generates false recognition but proposed scheme is dexterous enough to identify the TB infection automatically with very little execution instant and high accuracy. A learning model- Support Vector Machine (SVM) is useful for classifying impairment lungs. Before classifying the cells, image requires some enhancements i.e. Adaptive Histogram Equalization and Embossing that directionally compute the color differences. The novel system has been tested with 286 lung CT scan samples and on the basis of correct prediction; the system accuracy is 96.50%.