TY  - JOUR
AU  - Debata, Biswaranjan 
AU  - Priyadarshini, Rojalina 
AU  - Mohapatra, Sudhir Kumar 
AU  - Bedane, Tarikwa Tesfa 
PY  - 2025
TI  - Classification of X-Ray Images Using Convolutional Neural Network and Automatic Hyper-Parameter Selection to Detect Tuberculosis (TB)
JF  - Journal of Computer Science
VL  - 21
IS  - 2
DO  - 10.3844/jcssp.2025.413.423
UR  - https://thescipub.com/abstract/jcssp.2025.413.423
AB  - Tuberculosis (TB) is a major public health issue in India, contributing significantly to the global burden of respiratory diseases. This study introduces a Convolutional Neural Network (CNN)--based model for the early and cost-effective detection of TB using chest X-ray images. The proposed model, featuring 13 layers and automated hyperparameter selection, classifies images as infected or not infected. It is evaluated on three open datasets: Chest X-ray Masks and Labels, Tuberculosis X-ray (TB ×11 K), and Shenzhen. The model achieves an accuracy of 99.42% on the chest X-ray masks and label dataset, 99.27% on the TB ×11 K dataset, and 97.73% on the Shenzhen dataset, outperforming six existing models in terms of F1 score and precision. Unlike existing models that are tested on a single dataset, our model demonstrates consistent and robust performance across multiple datasets, highlighting its generalizability.