Detection of Covid-19 using Deep Learning Techniques

Abstract

The text discusses the global impact of infectious diseases, emphasizing COVID-19 as a significant example. It proposes using medical imaging (such as Chest X-ray and CT scans) and deep transfer learning to detect COVID-19 cases. The methodology focuses on employing pre-trained deep neural networks (ResNet50, InceptionV3, VGGNet-19, and Xception) along with data augmentation techniques. The study reveals that VGGNet-19 performs best when considering CT scans, while the refined Xception model excels with CXR images, achieving high precision, recall, F1-score, and accuracy values. When combining both modalities, VGG-19 presents the most favorable overall scores, showcasing the potential to automate the analysis of chest CT scans and X-ray images with high accuracy, particularly useful in situations with limited RT-PCR testing and resources.