Tuberculosis Diagnosis from Chest X-Ray Image Using Deep Learning Techniques

Tuberculosis (TB) is a significant global health challenge, necessitating precise and efficient diagnostic tools. This work investigates the effectiveness of various deep learning models such as EfficientNetB3, DenseNet121, VGG19, DenseNet201, and ResNet50 in detecting TB from chest X-ray images. Utilizing the TB Chest X-ray Database, we evaluated pre-trained models, developed custom architectures, and employed the k-fold validation method. The EfficientNetB3, with a customized design, demonstrated superior performance, while the DenseNet201 excelled in the k-fold validation setup. Our findings highlight the potential of the EfficientNetB3 and DenseNet201 in advancing TB diagnostics, offering significant improvements in speed and accuracy. We got 100% accuracy in EfficientnetB3. This study not only contributes to the growing field of AI in medical imaging but also underscores the ethical considerations required in deploying AI for healthcare, ensuring patient privacy and equitable algorithmic decisions. The outcomes of this study hold promise for improving real-time healthcare through more effective TB detection and patient care.