ABSTRACT:Millions of reported cases and associated deaths highlight the annual global threat posed by Tuberculosis (TB). Added to this, limited diagnostic services, particularly in rural Nigeria, worsen the prevalence of TB in the country. To address these challenges, this research explores the deployment of the deep learning model DenseNet121 to automate TB diagnosis from chest X-rays in low-resource settings such as Nigeria. The model aims to facilitate earlier TB detection in communities with inadequate access to diagnostic services. The absence of qualified TB radiologists in these communities further enhances the model’s potential. Based on analysis of a database comprising 4,200 chest X-ray images, the model achieved the following diagnostic metrics: 97.14% accuracy, 0.94 precision, 0.93 recall, and 0.90 F1 score. Such results provide sufficient evidence that the model will significantly improve the timely diagnosis and detection of TB cases. This illustrates the power of Artificial Intelligence tools in constraining and limited environments.