<![CDATA[Tuberculosis (TB) remains a significant global health challenge, requiring rapid and accurate diagnostic tools to prevent transmission. Chest X-ray (CXR) imaging is a primary screening method, yet manual interpretation is often subjective and prone to inconsistency. This study proposes an efficient automated detection framework using the EfficientNet-B0 architecture integrated with Contrast Limited Adaptive Histogram Equalization (CLAHE). The research utilizes the Shenzhen Dataset, employing CLAHE to enhance the visibility of pulmonary features by mitigating non-uniform illumination in radiographs. The model was modified with a Global Average Pooling (GAP) layer and a 0.5 dropout rate to optimize performance for binary classification. Experimental results demonstrate that the proposed framework achieved an Accuracy of 87.21%, a Sensitivity of 89.70%, and an Area Under the Curve (AUC) of 0.9350. Furthermore, the model exhibits high computational efficiency with a compact size of 20.5 MB and only 5.3 million parameters, significantly outperforming heavier architectures like ResNet-50. This study concludes that the combination of CLAHE-based enhancement and EfficientNet-B0 provides a robust and lightweight solution for TB screening, particularly suitable for deployment in resource-constrained clinical environments.]]>
Copyrights © 2026
