A Novel Approach to Detect Coronavirus Based on Lung x-ray Image and Convolutional Neural Networks

                                                                        Huda Mubarak Ismail^[1]1, Pedram Salehpour², Seyed Hadi Aghdasi Alamdari³

                                                                                                      ¹Ministry of Defense, Baghdad, Iraq 

                                                                       ^2,3Department of Electrical and Computer Engineering, University of Tabriz, Tabriz, Iran                                                                                                                

                                                                         Emails: hudamod2017@gmail.com ; Psalehpoor@tabrizu.ac.ir; aghdasi@tabrizu.ac.it

Abstract

The coronavirus has become a global crisis in recent months. The virus has disrupted or shut down many social, economic, sports, scientific, etc. activities. In addition to the medical importance of this Disease, its rapid and accurate diagnosis is an important need. In this study, we proposed a novel method to detect coronavirus using machine learning and classification algorithms based on lung images. In general, the method consists of two steps. At the first step, a convolutional neural network is trained with a data set of lung images that determine whether a viral infection exists or not. In the second step, another network is used to detect if the existence of viral infection is considered coronavirus or not. Experimental tests have been conducted that show the correct diagnosis can be made with 95.8%.

Keywords: Corona diagnosis; deep learning; convolutional neural networks; lung images.