Detection of Pneumonia Clouds From Chest X-ray Images

The use of an automated system in the X-ray image processing is very important to radiologists. To end this, many approaches for detecting lung diseases have been developed, and many imaging techniques such as computerized tomography (CT) and chest radiography (X-ray) have been employed. In this paper, an algorithm to detect pneumonia the chest X-ray is considered. Pneumonia, inflammatory lung disease appears as white patches of cloud on the chest X-ray images. On chest X-ray images, an automated system has been implemented using both with pathology and without pathology using Kaggle dataset. This paper aims to help efficiently detect pneumonia by segmenting the lung area and by extracting features. The lung area is segmented using the watershed algorithm and graph cut process. Gray Level Co-Occurrence Matrix (GLCM) features are extracted and fed to classifier for identifying the diseased lung from the healthy lung. The classification accuracy of 92% is accomplished by K-Nearest Neighbor (K-NN) classifier. The system has been tested with 300 chest X-ray images and has achieved sensitivity of 89%, specificity of 93% in classifying the lung diseases.