Pneumonia Classification Model using Deep Learning Algorithm

Abstract

The bacteria Streptococcus pneumoniae is the cause of pneumonia, a potentially fatal infectious disease that affects one or both lungs in humans. According to the World Health Organization (WHO), pneumonia is to blame for one in every three fatalities in India. Three classification categories are considered in this paper: Healthy, Viral and Bacterial infection. Chest X-rays that are used to diagnose pneumonia and must be evaluated by experienced radiotherapists in the medical sector. By combining three different classification techniques, a new hybrid Convolutional Neural Network (CNN) model is suggested in this regard. To classify CXR images, the first classification method makes use of Fully-Connected (FC) layers. The weights that result in the highest level of classification accuracy are retained after this model has been trained over a number of epochs. In the second method of classification, Machine Learning (ML) classifiers are used to classify the images, and the trained optimized weights are used to extract the features that are the most representative of CXR images. The proposed classifiers are used in an ensemble in the third classification method to classify CXR images. With an accuracy of 98.55 percent, the outcomes demonstrate that the proposed ensemble classifier, which combines Support Vector Machine (SVM), and other classifiers which performs the best. Finally, this model is used to create a Computer Automated Detection system that radiologists can use to accurately detect pneumonia.