Diagnosis of COVID-19 in CT images based on convolutional neural network (CNN)

Abstract

A new “coronavirus disease-2019 (COVID-19)” has spread quickly as an acute respiratory distress syndrome (ARDS) among individuals worldwide. Moreover, the number of COVID-19 test kits available in hospitals is limited compared to the growing number of cases every day. Therefore, It is necessary to introduce an automatic detection system as a fast alternative diagnostic method to prevent the spread of COVID-19 among humans. The purpose of this study is to propose an automated method using a machine learning method (Convolutional Neural Network (SimpNet model)) for the identification of COVID-19 pneumonia-infected patients using chest computed tomography (CT) images. Threshold and mathematical morphology were used to segment lung tissue as a region of interest (ROI). The Convolutional Neural Network (CNN) based on multi - Image augmentation technique was applied as a deep feature extraction technique and to identify CT samples with Covid 19 and Non-Covid 19. Specificity, Sensitivity, Accuracy, F1- score, Area Under Curve (AUC) were used as criteria to estimate the classification’s efficiency. The highest classification accuracy was achieved as 98.67%.

Ähnliche Arbeiten