Automated diagnosis of tuberculosis using deep learning techniques

In recent decades, automatic diagnosis using machine-learning techniques have been the focus of research. The ability of classifiers to be able to perform a specific task as good as a trained profession opens the door to a lot of applications. Mycobacterium Tuberculosis (TB) is a deadly disease that plagues most developing countries due to weak health system infrastructure that can be able to quickly diagnose and treat the disease. The World Health Organisation (WHO) has set years 2030 and 2035 as milestones for significant reduction in new infections and deaths. The WHO reports that lack of well-trained professionals to accurately diagnose TB and insufficient or fragile public health systems which are mostly overwhelmed are the major factors that have slowed the eradication of the TB endemic. Convolutional neural networks (CNNs) have demonstrated remarkable success in the field of computer vision i.e. image recognition and detection. Consequently, common methodology for detecting TB is through radiology combined with previous success CNN have achieved in image classification makes them the perfect candidate to classify Chest X-Ray (CXR) images of potential TB patients. In this study, we propose three types of CNN trained using two public datasets and another which was collected from Konya Education and Research Hospital, Konya, Turkey. Also, the CNN architectures were integrated an extra layer called Spatial Pyramid Pooling (SPP) a methodology that equips convolutional neural networks with the ability for robust feature pooling by using spatial bins. Our results indicate a huge potential for an automated system to diagnose tuberculosis with accuracies above a radiologist professional. Keywords: Convolutional Neural Network, Deep Learning, Tuberculosis, Spatial Pyramid Pooling.