COVID-19 Chest X-Ray Classification with Augmented GAN

Abstract

AbstractCoronavirus disease (COVID-19) is a major concern now. According to the Globe Health Organization, the coronavirus (COVID-19) epidemic is straining healthcare systems worldwide (WHO). Early-stage detection using artificial intelligence of this virus will help in the fast recovery. Early identification of this infection utilizing artificial intelligence will aid in its quick recovery. In the fight against COVID-19, it’s critical to have a positive chest X-ray for infected patients. Early research suggests that chest X-ray abnormalities in COVID-19 patients are common. Using augmented chest X-ray images, this research proposes a novel model for identifying the presence of COVID-19. As medical images are sensitive, GAN (Generative Adversarial Networks) augments chest X-Ray images. Augmented images are classified through our proposed model. Then classified images are segmented for validation. GAN augmentation model consists of a generator and a discriminator. Convolutional neural networks (CNNs) based classification model needs a substantial amount of training data. Augmentation increased dataset into 10x amount. Before augmentation, our initial model received 94.42% accuracy, and after augmentation, our final model accuracy raised to 98.58%. We hope our model will detect COVID-19 presence accurately through X-Ray images.KeywordsChest X-rayClassificationAugmentationGANSegmentationCOVID-19Artificial intelligence