An accurate neural network algorithm to diagnose Covid-19 from CT images

Abstract

Background: A new coronavirus appeared in late December 2019 in Wuhan, China. He was named Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2). This virus is responsible for Covid-19, the name given to the disease associated with it. It spreads worldwide, infecting more than a million people and killing more than 70 miles. The rapid and accurate diagnosis of suspected Covid-19 cases plays a crucial role in medical treatment and timely quarantine. Materials and Methods: In order to counter the Covid-19 pandemic, we have developed a method for the automatic detection of Covid -19, from 2D computed tomography (CT) chest images. It is a supervised software system based on the ANN (Artificial Neural Network) algorithm. Pulmonary CT images were collected from multiple international datasets, with a total of 395 images: 70% were used for training and 30% were used for testing. For each patient, the lungs were segmented using simple thresholding. Then, the segmented lungs were fed into a neural network to predict the probability of SARS-CoV-2 infectious. Results: The internal validation achieved a total accuracy of 97.5% with a specificity of 96.6 % and a 100 % sensitivity. Conclusion: These results demonstrate the proof-ofprinciple for using artificial intelligence to extract radiological features for timely and accurate Covid-19 diagnosis.