Pneumonia Detection on Chest X-Ray Using Machine Learning Paradigm

The chest radiograph is the globally accepted standard used for analysis of pulmonary diseases. This paper presents a method for automatic detection of pneumonia on segmented lungs using machine learning paradigm. The paper focuses on pixels in lungs segmented ROI (Region of Interest) that are more contributing toward pneumonia detection than the surrounding regions, thus the features of lungs segmented ROI confined area is extracted. The proposed method has been examined using five benchmarked classifiers named Multilayer Perceptron, Random forest, Sequential Minimal Optimization (SMO), Logistic Regression, and Classification via Regression. A dataset of a total of 412 chest X-ray images containing 206 normal and 206 pneumonic cases from the ChestX-ray14 dataset are used in experiments. The performance of the proposed method is compared with the traditional method using benchmarked classifiers. Experimental results demonstrate that the proposed method outperformed the existing method attaining a significantly higher accuracy of 95.63% with the Logistic Regression classifier and 95.39% with Multilayer Perceptron.