Detection of Tuberculosis using Hybrid Features from Chest Radiographs

  • Maham Mehr Awan Department of Software Engineering, Lahore Garrison University, LGU
  • Ayesha Fatima Department of Computer Engineering, National University of Science and Technology, NUST
  • Kinza Mehr Awan Department of Software Engineering, University of Management and Technology, UMT
Keywords: Keywords: Tuberculosis (TB), Chest Radiograph (CXR), Classification, Computer-Aided Diagnosis (CAD).


Tuberculosis is a contagious disease, but it’s diagnosis is still a difficult and challenging task as it is considered a big threat everywhere on the planet. Literature shows that underdeveloped countries widely use chest radiographs (X ray images) for the diagnosis of tuberculosis. Low accuracy of results and high cost are the two main reasons due to which most of the available methods are not useful for radiologists. In our research, we proposed a detection technique in which features extraction is performed on the basis of their texture, intensity and shape. For evaluating the performance of our proposed methodology, Montgomery Country (MC) dataset is used. It is a publically available data set which consists of 138 CXRs; among them, 80 CXRs are normal and 58 CXRs are malignant. The results of the proposed technique have outperformed state of the art methodologies on the MC dataset as it has shown 81.16% accuracy.

How to Cite
Maham Mehr Awan, Ayesha Fatima, & Kinza Mehr Awan. (2020). Detection of Tuberculosis using Hybrid Features from Chest Radiographs. Lahore Garrison University Research Journal of Computer Science and Information Technology, 4(3), 39-46.