CXR images versus CT images for detection and analysis of COVID-19 using deep learning techniques

Early detection is essential to fight against the spread COVID-19 and other pulmonary diseases. Novel studies show the capability of deep learning approaches to detect COVID-19 in x-ray images. This chapter presents an exhaustive analysis between two kinds of x-ray–based images, computed tomography (CT) and chest x-ray (CXR) to contrast their limitations and advantages applied in deep learning techniques. To address this study, two different datasets were collected, CT dataset composed of 14 576 slices, and CXR dataset with 7216 images. Subsequently, a complete framework was developed to evaluate the CT dataset and the CXR dataset in three different CNN architectures (VGG19, DenseNet, and ResNet) for diagnosing COVID-19. Evaluation results presented show the feasibility and explore the differences between our proposals.