Hybrid Classification Model by Using CNN to Detect COVID-19 and Chest Related Problems

Abstract

The COVID-19 pandemic has created havoc on the lives of many people and their health all over the world. It has been increasing very rapidly, one must find an effective model/method to detect COVID-19 in order to help the Health Care System. Chest X-ray is one of the reliable diagnostic technologies, which helps in the identification of COVID-19. Despite the fact that there are numerous deep learning methodologies for identifying COVID -19, these methodologies are useless if they only detect one type of illness while ignoring the others. This study proposed a Hybrid Classification model based on CNN (Convolutional Neural Network) for more efficient detection of COVID-19 from Chest X-Rays. Using CNN, this study differentiates COVID-19 affected chest X-Ray images from normal chest X-Ray images and eight additional chest disorders (Cardiomegaly, Atelectasis, Infiltration, Effusion, Nodule, Pneumonia, Mass, Pneumothorax). The Hybrid Classification Model contains two classifiers, Classifier-1 and Classifier-2. In Classifier-1, it contains the information about Normal Chest X-rays images and chest X-ray images that have been affected by COVID-19 and whereas in the Classifier-2, it contains the information about other 8 chest diseases. For getting highest accuracy of Classifier-1 and Classifier-2 models, this research work utilizes several models i.e., ResNet50, InceptionResNetV2, VGG16, DensNet121 and Mobile Net. Based on all these models, this research work considers ResNet50 for Classifier-1, and DensNet121 for Classifier-2, Because these two models had given the highest accuracy compared to other models.

Ähnliche Arbeiten