Enhancing Chest X-Ray Image Classification for Lung Diseases through Machine Learning

Abstract

Chest X-ray imaging is a widely used diagnostic tool for the detection and classification of various lung diseases. In this study, we propose a methodology to enhance the classification accuracy of chest X-ray images by leveraging machine learning techniques. Specifically, we employ the VGG-19 architecture to classify chest X-ray images into five distinct lung conditions: normal, tuberculosis, COVID-19, lung opacity, and pneumonia. A comprehensive dataset consisting of 31,787 chest X-ray images sourced from multiple medical institutions and hospitals worldwide is collected. Each image is labeled by expert radiologists with one of the five lung conditions. The dataset is then preprocessed and trained with 30 epochs. The trained VGGnet model achieved overall test accuracy of 95.11%, demonstrating its capability to accurately classify chest X-ray images into the five targeted lung conditions. The proposed methodology holds significant potential for improving the efficiency and accuracy of lung disease diagnosis based on chest X-ray images. By employing machine learning techniques, we can automate the classification process, providing clinicians with valuable decision support and expediting treatment plans. Moreover, the developed model can assist in the identification of lung diseases, including critical conditions such as COVID-19, enabling prompt and effective patient management.