An Efficient Method to Predict Pneumonia from Chest X-Rays Using Deep Learning Approach
Uzair Shah, Alaa Abd-Alrazeq, Tanvir Alam, Mowafa Househ, Zubair Shah
Abstract
Pneumonia is a severe health problem causing millions of deaths every year. The aim of this study was to develop an advanced deep learning-based architecture to detect pneumonia using chest X-ray images. We utilized a convolutional neural network (CNN) based on VGG16 architecture consisting of 16 fully connected convolutional layers. A total of 5856 high-resolution frontal view chest X-ray images were used for training, validating, and testing the model. The model achieved an accuracy of 96.6%, sensitivity of 98.1%, specificity of 92.4%, precision of 97.2%, and a F1 Score of 97.6%. This indicates that the model has an excellent performance in classifying pneumonia cases and normal cases. We believe, the proposed model will reduce physician workload, expand the performance of pneumonia screening programs, and improve healthcare service.