<![CDATA[Tuberculosis (TB) is an infectious disease that poses a global health threat. Early diagnosis through chest X-ray (CXR) imaging is effective in reducing transmission and improving patient recovery rates. However, the limited number of radiologists in high TB burden areas hampers rapid and accurate detection. This study aims to improve TB diagnosis accuracy using deep learning models. Convolutional Neural Networks (CNN) are applied to analyze CXR images to support automated detection in regions with limited radiology personnel. The method involves image processing using Contrast Limited Adaptive Histogram Equalization (CLAHE) to enhance image quality. A public dataset consisting of 2,188 images was used, with preprocessing steps including resizing, normalization, and augmentation. The DenseNet201 model was employed as the main architecture, trained for 10 epochs with various batch sizes to evaluate its performance. Results show that the combination of CLAHE and DenseNet201 achieved the highest accuracy of 94.84%. Image quality enhancement with CLAHE proved to improve accuracy compared to models without preprocessing. This research contributes to enhancing the efficiency of automated early TB detection, reducing reliance on radiologists, and accelerating clinical decision-making.]]>
Copyrights © 2025
