An Automated System for the Classification of COVID-19, Suspected COVID-19 and Healthy Lung CT Images based on Local Binary Pattern and Deep Learning Features

Abstract

Because of the inadequate capacity and a substantial surge of probable COVID-19 cases, several health systems around worldwide have collapsed. As a result, the requirement for a rapid, effective, and precise way to reduce radiologists' workload in diagnosing suspected instances has arisen. The goal of the present study is to develop a novel system to automatically diagnose and classify lung CT scans into three categories: suspected covid-19, covid-19, and healthy lung scans. Before feature extraction using convolutional neural network (CNN) and Local Binary Pattern (LBP) approaches, the CT scans are first pre-processed through implementing a set of algorithms. Lastly, with the use of the support vector machine (SVM) model, such features are divided into three groups. The maximum accuracy attained in classifying a dataset of 351 CT scans of the lungs was 98.22%. The outcomes of the experiments show that merging the extracted features increases the effectiveness of lung classification CT scans.