Paper Title
Automatic and Interpretable Diagnosis Of Covid-19 Using convolutional Neural Networks

Abstract
COVID-19 has killed more than 435,000 people and infected over 7.94 million people globally. This number is continuing to grow every day. Currently, many hospitals are facing severe challenges because of the influx of patients with critical conditions. It takes labs approximately 24 hours to run COVID-19 tests; however, it can take several days for patients to receive their results. Furthermore, there is a need for automated toolkits that can be used in remote areas to provide an accurate, low-cost diagnosis.Currently, 400 million people lack access to healthcare services. The engineering goal of this project was to create an automated system with diagnostic accuracy comparable to radiologists with a faster diagnostic rate. The system would also be incorporated with diagnostic interpretability to assist radiologists and minimize false-positive and false-negative diagnoses by providing contributory features. The system would also be assistive in remote areas where residents don't have access to specialized physicians. CNNs have become an increasingly powerful imaging tool in the field of machine learning. These neural networks provide promising results in medical imaging. In this study, a new model for interpretable and automatic COVID-19 diagnosis is proposed. The model used X-ray images for training and testing and produced a classification accuracy of 93%. Furthermore, Chest-X rays are classified as either COVID or non-COVID. The proposed model can be incorporated in PACS, which is a widely used tool in healthcare to store images and reports, to validate radiologist screenings. The model can also be employed in remote areas as an efficient telemedicine tool. Keywords - CXR (chest x-ray), CNN (convolutional neural networks), deep learning