Deep Learning Based Tuberculosis and Pneumonia Disease Detection Using CNN

Tuberculosis and pneumonia are debilitating respiratory diseases that pose significant health challenges around the world. For these disorders to be treated and controlled effectively, a timely and precise diagnosis is essential. The aforementioned advancements have resulted in the emergence of novel technologies utilized for the identification and assessment of TB and pneumonia. These technologies encompass deep-learning models and computer-aided diagnosis systems. Our study aims to create an automated method to diagnose tuberculosis and pneumonia with one command. In this investigation, we have curated a comprehensive dataset containing chest X-ray images with three distinct classes: normal, tuberculosis-infected, and pneumonia-infected cases. We have implemented deep learning models, VGG16, ResNet50, MobileNet, and DenseNet, including InceptionV3 to classify these image diseases. Our results show that InceptionV3 achieved the highest accuracy of 99.24% in disease detection. The utilization of this method can aid healthcare professionals in the prompt and precise identification of TB and pneumonia, hence enhancing patient outcomes and alleviating the strain on healthcare systems.