Artificial Intelligence System for Classification of COVID-19 from CXR Images

Abstract

Coronavirus Disease 2019 (COVID-19) becomes the crucial disease in recent times. Further, many variants of COVID-19 are evolving from the broad family of severe acute respiratory syndrome (SARS). Thus, the detection of all these variants by using Real-time polymerase chain reaction (RT-PCR) test is a difficult task and time taking. In addition, the conventional methods are failed to classify the COVID-19 in early stage due to complex architecture of chest x-ray (CXR) image. Therefore, this article is focused on implementation of deep learning convolutional neural network (DLCNN) based artificial intelligence approach for classifying COVID-19 disease. Initially, the hybrid features are extracted from CXR dataset by using Multi Block Local Binary Pattern (MB-LBP), and Weber local descriptor (WLD). Further, increment component analysis (ICA) is used to reduce features, which generates best features. Then, DLCNN model is trained with these features for classification of COVID-19 for each test CXR image. The simulation results show that proposed classification resulted in better subjective and object performance as compared to conventional machine learning and deep learning methods.