Automated detection of multiple lung diseases in X-ray images using deep learning techniques

Imaging studies, and particularly X-ray examinations, are among the most important components in the early diagnosis of many diseases. This paper proposes the use of a framework based on the convolutional neural networks (CNNs) to determine the presence of diseases like COVID-19, Pneumonia and Tuberculosis. In addition, the proposed work is able to recall complex patterns by simply employing the use of numerous datasets and transfer learning techniques. The ResNet model is trained for 65 epochs and attained the highest accuracy of 0.97, MobileNet, VGG Net and Google Net are achieved an accuracy of 0.9558, 0.9388, and 0.92, respectively. However, the comparison of performance with traditional diagnostic techniques using sensitivity, specificity, and accuracy shows that the proposed work is much better, and fit for early diagnosis.