Deep Learning-based Model for the Classification of Tuberculosis using Convolution Neural Network

Mycobacterium tuberculosis is a bacterium which generates tuberculosis disease (TB). It is an extremely contagious bacterial infection which may result in death. Tuberculosis is one of the world’s top 10 killer diseases. A six-month course of antibiotics is the standard treatment for tuberculosis, which places a consequential burden on both patients and medical systems around the world. Fortunately, most patients who receive treatment for just two to three months recover fully and don’t need any additional care. However, when the stage advances, it becomes very difficult to cure. Hence, it is required to early diagnose TB to reduce the mortality rate and cure the disease in the early stages. Nowadays, for the detection of TB, image-analysis-based computer-assisted diagnostic methods are used. In this work, a convolutional neural network (CNN) model has been employed to predict TB disease which has achieved the prediction outcome with an accuracy of 99.41% and a loss rate of 0.0284. Additionally, a comparison analysis with the state-of-the-art has also been carried out in order to validate the proposed model’s results. The findings indicate that early detection of TB can significantly decrease the mortality rate and enable timely prediction of the diseases.