User Friendly Web Based Deep Learning System for TB and Pneumonia Diagnosis in Developing Countries

Tuberculosis (TB) and pneumonia are two of the leading causes of death and disability worldwide. Both diseases are preventable and treatable, but early diagnosis and treatment are crucial, especially in developing countries. Chest X-ray (CXR) is the most widely used imaging modality for diagnosing TB and pneumonia, but it is time-consuming and subjective to interpret. Deep learning is a subfield within the domain of machine learning that uses artificial neural networks as its primary computational framework for acquiring knowledge from data. Deep learning models have demonstrated efficacy in a range of medical imaging applications, encompassing the accurate detection of tuberculosis (TB) and pneumonia. This paper proposes a deep learning-based system for accurately and efficiently diagnosing TB and pneumonia from CXR images using the VGG19 architecture. The system was trained and evaluated on a large dataset of CXR images from patients with TB, pneumonia, and normal cases, achieving an accuracy of 99%. The authors also evaluated the performance of eight different deep learning algorithms for the classification of normal and abnormal CXR images. The VGG19 algorithm achieved the highest accuracy (99%), followed by DenseNet121 (98%) and Inception V3 (97%). The system is user-friendly and accessible through a web interface, making it accessible to healthcare professionals in all settings, including developing countries. The suggested method has the potential to greatly enhance TB and pneumonia detection and treatment., especially in developing countries. By automating the image analysis process and improving diagnosis accuracy, the system can help reduce mortality and morbidity associated with these diseases