A Deep Learning Framework for the Classification of Lung Diseases Using Chest X-Ray Images

Abstract

AbstractAutomated screening and classification of various lesions in medical images can assist clinicians in the treatment and management of many systemic and localized diseases. Manual inspection of medical images is often expensive and time-consuming. Automatic image-analysis employing computers can alleviate the difficulties of manual methods for screening a large amount of generated images. Inspired by the great success of deep learning, we propose a diagnostic system that can classify various lung diseases from chest X-ray images. In this work, chest X-ray images are applied to a deep-learning algorithm for classifying images into pneumothorax, viral pneumonia, COVID-19 pneumonia and healthy cases. The proposed system is trained with a set of 4731 chest X-ray images, and obtained an overall classification accuracy of 99% in images taken from two publicly available data sets. The promising results demonstrate the proposed system’s effectiveness as a diagnostic tool to assist health care professionals for categorizing images in any of the four classes.KeywordsCNNPneumothoraxViral pneumoniaCOVID-19