An Effective Approach for Early Diagnosis of Tuberculosis using Deep Learning Technique

Abstract

Tuberculosis (TB) is a communicable disease that mostly infects the lungs. The tuberculosis is generated by the bacteria called Mycobacterium Tuberculosis (MTb). It easily outspreads via the air when people who have active tuberculosis spit, cough or talk. The diagnosis of tuberculosis at an early stage is a tedious process. If the bacilli are identified at an earlier stage than it is trouble-free to provide treatment. The presence of tuberculosis in human beings is being found using Chest X-ray image, Sputum image and employing technologies are Computer-Aided Detection, Feature Selection, Neural Network, Active contour. Even though several methods are employed, no single method gives more accuracy in diagnosing TB at earlier stage. This leads to enormous death all over the world. Therefore an efficient system is implemented for the early diagnosis of tuberculosis bacteria using U-Net in Deep Learning method. This technology predominate the existing methods efficiently for early diagnosis of TB, thereby death rate can be reduced and prevented.