Leveraging Deep Learning Techniques for Chest X-Ray Tuberculosis Detection

View at publisher Summarise with VIRAT
Authors: Shashikiran S, Srinivas Babu N, Kirana B V, S V Darshan Patil, Pramod R, Shravan Kumar Gogi
Cited by: 0
References: 15

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Despite the availability of highly advanced therapies, tuberculosis (TB) remains a serious global health concern that affects millions of people due to a lack of technological advancements in detection and treatment. Decimating the spread and improving treatment outcomes requires accurate identification and early detection. Because they are performed by lab workers, traditional diagnostic techniques like sputum microscopy and culture are labor-intensive and prone to human error. Deep learning (DL) has made significant strides recently, offering considerable potential for increasing and automating diagnosis accuracy. Our study suggests a DL-based method for identifying tuberculosis in chest X-rays. Our model surpasses conventional techniques with an astounding accuracy of 97.32% and a loss of 8.19% after training on a big dataset. The model efficiently recognizes TB-related patterns with low false positives by applying convolutional neural networks (CNN) as a base, combined with transfer learning techniques (SqueezeNet) and explainable AI techniques like Grad-CAM. This method has the potential to transform tuberculosis diagnosis and provide more dependable, scalable, and fast solutions to global healthcare systems.

Fields of Study & Subjects

COVID-19 diagnosis using AI AI in cancer detection Tuberculosis Research and Epidemiology

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1400–1404
Publication date
21 January 2026
Language
EN