PCNNTB: Pre-Activated Deep Convolutional Neural Networks Based Tuberculosis Severity Classification

Tuberculosis (TB) is a infection disease and mainly it affects the human lungs. As stated by the World Health Organization, 10.6 million population of the whole world is affected with Tuberculosis. If we diagnose TB at an early stage, then we can control this disease. Tuberculosis is a major challenge problem for immunocompromised HIV/AIDS patients. In this paper, we proposes a novel technique to automatically detect and classify the Tuberculosis using the Preactivated Deep Convolutional Neural Networks Based Tuberculosis Learner method (PCNNTB). The proposed deep learning model provides more discriminate information to more accurately classify the positive or negative images. In this work, the performance of the proposed method is evaluated on the standard ImageClef 2019 dataset. The experimental results demonstrate the proposed method provides promising results on ImageClef 2019 challenging dataset.