Improving Tuberculosis Detection with Deep Learning in X-Ray Imaging

Abstract

Background and Purpose: Tuberculosis (TB) continues to pose a major global health burden, particularly in low-resource settings where timely and accurate diagnosis remains a challenge. Conventional diagnostic methods such as sputum smear microscopy and bacterial culture are often slow, labor-intensive, and susceptible to human error. While several studies have explored alternative diagnostic tools, there remains a critical need for efficient, accurate, and scalable solutions. This study aimed to address this gap by developing a deep learning (DL)-based diagnostic model using chest X-ray images to detect TB with high precision and speed.

Methods: A Convolutional Neural Network (CNN) model based on the VGG19 architecture was developed and trained on a dataset of labeled chest X-ray images. The model was optimized to identify radiographic features indicative of TB. Performance metrics including precision, recall, F1-score, and overall accuracy were used to evaluate the model’s diagnostic effectiveness.

Results: The proposed DL model demonstrated strong diagnostic performance, achieving a precision of 96%, recall of 96%, F1-score of 96%, and an overall accuracy of 96%. These results indicate the model’s robustness in identifying TB cases from chest X-ray images with minimal false positives and false negatives.

Conclusions: The findings underscore the potential of AI-driven diagnostic tools to significantly enhance TB detection, particularly in settings with limited access to laboratory infrastructure. The VGG19-based model offers a promising pathway toward faster, more reliable, and scalable TB diagnosis, contributing to improved disease management and control efforts globally.