Abstract: Covid-19 (Coronavirus Disease) is an infectious disease caused by a group of viruses named Coronavirus. It primarily affects the lungs of a person. With millions of death across the globe due to this virus, it is necessary to create a system that predicts whether a person is infected with the virus. The use of statistical and machine learning algorithms can be used for the early prediction of the disease. In this paper, a Deep Convolutional Neural Network(DCNN) has been used to predict whether a pair of lungs are infected with Covid-19 or Healthy or just Viral Pneumonia. Covid-19 Radiography dataset has been used to develop the CNN model to predict the presence of the disease in the lungs. In pre-processing, initially, the image data were normalized and feature extraction was carried out for enhanced accuracy. Finally, data augmentation increases the number of images that improves the accuracy of the proposed model. This paper discusses the classification accuracy, precision, recall, f-score and AUC score which are used as evaluation metrics for the proposed DCNN model. The performance of the proposed DCNN is compared with that of two pre-trained models: ResNet-50 and Inception-V3. The proposed model achieved a training accuracy of 98.86% and test accuracy of 94%. This project can help in the early prediction of the disease by using x-ray images of the lungs.
Keywords: Covid-19, Prediction, Convolution Neural Network, Accuracy, Radiograph, Classification
|
DOI:
10.17148/IJARCCE.2021.101024
