Abstract:The outbreak of corona virus disease in December 2019 in China spread rapidly across all parts of the world by January 2020. The World Health Organization (WHO) termed it as COVID-19 and declared it a pandemic on January 30, 2020. Till June 8th, 2020, the number of confirmed cases is around 7 million globally, and the global fatality rate is around 3-4%. Since it is a highly contagious disease and is spreading rapidly, governments of almost all of the affected countries are taking it on priority to isolate infected individuals as early as possible. The general symptoms of COVID-19 patients are flu-like such as fever, cough, dyspnea, breathing problem, and viral pneumonia. But these symptoms alone are not significant. There are many cases where individuals are asymptomatic but their chest CT scan and the pathogenic test were COVID-19 positive. So, along with symptoms, positive pathogenic testing and positive CT images/X-Rays of the chest are being used to diagnose the disease. Deep Learning (DL) techniques specifically Convolutional Neural Networks (CNN) has proven successful in medical imaging classification. Four different deep CNN architectures were investigated on images of chest X-Rays for diagnosis of COVID-19. These models have been pre-trained on the train-test database thereby reducing the need for large training sets as they have pre-trained weights. It was observed that CNN based architectures have the potential for diagnosis of COVID-19 disease.
Keywords:Convolutional Neural Networks (CNN), COVID-19 positive, X-Rays for diagnosis of COVID-19, Deep Learning (DL), pneumonia or no pneumonia, Artificial Intelligence, RGB2GRAY, image pre-processing and resizing.
| DOI: 10.17148/IJARCCE.2020.91015