Automatic Detection of Tuberculosis Using Deep Learning Methods

Part of the Springer Proceedings in Business and Economics book series (SPBE)


In this paper, we present a deep learning based approach for automatically detecting tuberculosis manifestation from chest X-ray images. India is the country with the highest burden of tuberculosis. A chest radiograph in symptomatic patients is used to diagnose active tuberculosis. This screening method is ideally done at the primary health care centres where a clinician is available and sometimes through mobile X-ray unit. The major challenge for this method of screening is timely reporting and further follow-up of patient for initiation of treatment. We built multiple convolutional neural networks, the state-of-the-art deep learning algorithm, to build the model for automatic tuberculosis diagnosis. We classified the chest X-rays into two categories, namely, tuberculosis presence and tuberculosis absence. The dataset used to train the model contained 678 images, having 340 normal chest X-rays and 338 chest X-rays with tuberculosis manifestation. The validation dataset contained 235 images, which observed a sensitivity of 84.91% and a specificity of 93.02%. This demonstrates the potential of convolutional neural networks to automatically classify chest X-rays in real time.


Convolutional neural network Machine learning Artificial intelligence Pattern recognition Computer vision 


  1. Bengio, Y., Simard, P., & Frasconi, P. (1994). Learning long-term dependencies with gradient descent is difficult. IEEE Transactions on Neural Networks, 5(2), 157–166.CrossRefGoogle Scholar
  2. He, K., Zhang, X., Ren, S., & Sun, J. (October 2016a). Identity mappings in deep residual networks. In European Conference on Computer Vision (pp. 630–645). Springer International Publishing.Google Scholar
  3. He, K., Zhang, X., Ren, S., & Sun, J. (2016b). Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 770–778).Google Scholar
  4. Ioffe, S., & Szegedy, C. (June 2015). Batch normalization: Accelerating deep network training by reducing internal covariate shift. In International Conference on Machine Learning (pp. 448–456).Google Scholar
  5. Jaeger, S., Karargyris, A., Antani, S., & Thoma, G. (August 2012). Detecting tuberculosis in radiographs using combined lung masks. In Engineering in Medicine and Biology Society (EMBC), 2012 Annual International Conference of the IEEE (pp. 4978–4981). IEEE.Google Scholar
  6. Jaeger, S., Karargyris, A., Candemir, S., Folio, L., Siegelman, J., Callaghan, F., et al. (2014). Automatic tuberculosis screening using chest radiographs. IEEE Transactions on Medical Imaging33(2), 233–245.Google Scholar
  7. Krizhevsky, A., Sutskever, I., & Hinton, G. (2012) Imagenet classification with deep convolutional neural networks. In NIPS.Google Scholar
  8. LeCun, Y., Boser, B. E., Denker, J. S., Henderson, D., Howard, R. E., Hubbard, W. E., et al. (1990). Handwritten digit recognition with a back-propagation network. In Advances in Neural Information Processing Systems (pp. 396–404).Google Scholar
  9. Murtagh, K. A. T. H. L. E. E. N. (1980). Unreliability of the Mantoux test using 1 TU PPD in excluding childhood tuberculosis in Papua New Guinea. Archives of Disease in Childhood, 55(10), 795–799.CrossRefGoogle Scholar
  10. Perkins, M. D. (2000). New diagnostic tools for tuberculosis [The Eddie O’Brien Lecture]. The International Journal of Tuberculosis and Lung Disease, 4(12), S182–S188.Google Scholar
  11. Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., et al. (2015). Imagenet large scale visual recognition challenge. International Journal of Computer Vision115(3), 211–252.Google Scholar
  12. Simonyan, K., & Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv:1409.1556.
  13. Srivastava, N., Hinton, G. E., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. (2014). Dropout: a simple way to prevent neural networks from overfitting. Journal of Machine Learning Research, 15(1), 1929–1958.Google Scholar
  14. Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., et al. (2015). Going deeper with convolutions. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 1–9).Google Scholar
  15. Van Cleeff, M. R. A., Kivihya-Ndugga, L. E., Meme, H., Odhiambo, J. A., & Klatser, P. R. (2005). The role and performance of chest X-ray for the diagnosis of tuberculosis: a cost-effectiveness analysis in Nairobi. Kenya. BMC Infectious Diseases, 5(1), 111.CrossRefGoogle Scholar
  16. Van Ginneken, B., Katsuragawa, S., ter Haar Romeny, B. M., Doi, K., & Viergever, M. A. (2002). Automatic detection of abnormalities in chest radiographs using local texture analysis. IEEE Transactions on Medical Imaging, 21(2), 139–149.CrossRefGoogle Scholar
  17. World Health Organization, WHO (2016). Global Tuberculosis Report 2016.
  18. Yang, J., Yu, K., Gong, Y., & Huang, T. (June 2009). Linear spatial pyramid matching using sparse coding for image classification. In IEEE Conference on Computer Vision and Pattern Recognition, 2009. CVPR 2009. (pp. 1794–1801). IEEE.Google Scholar

Copyright information

© Springer Nature Singapore Pte Ltd. 2019

Authors and Affiliations

  1. 1.Enlightiks Business SolutionsBangaloreIndia

Personalised recommendations