A Hybrid Deep Neural approach for multi-class Classification of novel Corona Virus (COVID-19) using X-ray images

Abstract

People all around the world are facing challenges to survive due to Corona Virus (Covid-19). Pneumonia is often caused by COVID-19. Biomedical field has witnessed the success of Artificial Intelligence (AI) models for automatic diseases analyses and detection. Deep Learning (DL), a sub-field of AI, is used in this work to classify COVID-19 from Normal and Pneumonia patients. Three architectures i.e., Novel Convolutional Neural Network (N-CNN), Convolutional Neural Network- Long Short-Term Memory (CNN-LSTM) and Convolutional Neural Network-Random Forest (CNN-RF) models are proposed in this work for the classification of covid19 images from pneumonia and normal cases. We have used the X-ray image dataset in which 1212 training images consists of 404 images for each class and 300 validation images in which 100 images for each class. Five pre-trained models (VGG-19, VGG16, ResNet50, Inception v3 and Inceptio$\mathrm{n}_{-}$ResNetv2) are used to compare the classification performance with the proposed models. Among these pre-trained models and three proposed models, CNN-RF model outperformed and achieved an accuracy of 94.66% whereas N-CNN and CNN-LSTM models got an accuracy of 89.67% and 90.33% respectively.