Pulmonary tuberculosis (TB), caused by Mycobacterium tuberculosis, continues to pose a significant global health challenge, emphasizing the need for diagnostic tools that are both rapid and accurate. Chest X-rays are commonly used for initial screening, but their manual interpretation can be subjective and time-consuming. Traditional machine learning techniques, such as support vector machines (SVMs) and random forests, rely on handcrafted features and typically achieve moderate accuracy levels (between 80% and 90%). However, these methods often struggle with the variability of lesions and inconsistencies in image quality. In this study, we propose a deep learning-based approach for automated TB detection, built on the MONAI framework and the DenseNet121 architecture. Our methodology incorporates advanced preprocessing steps, including normalization and standardization, to enhance image quality and ensure optimal input for model training. Unlike traditional methods limited by shallow feature extraction, DenseNet121, a convolutional neural network known for its ability to learn deep, hierarchical features, demonstrated strong performance, reaching a test accuracy of 99.08%. All experiments were conducted on a carefully curated dataset, ensuring reliable and unbiased evaluation. By effectively handling variations in image quality and lesion presentation, our approach surpasses conventional techniques in accuracy and robustness. Thanks to its scalability and adaptability, the proposed model holds strong potential for integration into real-world clinical workflows. These results highlight the transformative role that deep learning can play in medical imaging, offering a powerful and scalable solution to enhance TB diagnosis and strengthen global public health initiatives.