Automated Tuberculosis Detection Using Convolutional Neural Networks on Chest X-Ray Images: A High-Accuracy Diagnostic Approach

Tuberculosis is one among the major health burdens across the globe, especially in developing regions. It is an infectious disease that is caused by Mycobacterium tuberculosis, generally affecting the lungs, though it may spread to other organs if not cured. Early diagnosis of TB is very important for receiving effective treatment and, as such, control the spread. In this work, the authors develop a CNN model to classify chest X-rays into two classes: TB-positive and TB-negative. The dataset had 4,200 labeled chest X-ray images. Of those, 3,500 were from patients without tuberculosis and 700 were from diagnosed TB patients. This dataset also trained the CNN and further validated it to result in an overall accuracy of 97%. The model performed with a precision of 0.99, a recall of 0.99, and F1-score of 0.99 for the TB-negative class, while for the TB-positive class the model performed at a precision of 0.96, recall of 0.94, and F1-score of 0.95. This points to very high performance of the model for both classes, with a weighted average F1-score of 0.98. These findings indeed show that CNN-based diagnosis for tuberculosis using X-ray images holds promise as an ancillary tool for healthcare professionals, particularly in resource-constrained settings. It also points to the potential of machine learning to improve diagnostic precision and reduce the burden of TB by way of early detection.