Use of Chest X-Rays in Predicting Pneumonia

Abstract

Pneumonia is a fatal infection that affects one or both lungs in humans and is often caused by Streptococcus pneumonia. The aim of the current study was to examine risk factors for death from pneumonia in young children. Therefore, implementing an autonomous pneumonia detection system would be beneficial, especially in remote areas, as it could save many lives and help prevent, treat and control the disease.
In this paper, different models like KNN, Random Forest, Decision Tree and CNN were trained from 5856 dataset images at 64 x 64 pixel resolutions. As KNN, Random Forest and Decision Tree are traditional models with around 70-80% accuracy, which are used widely, we intend to use CNN and hope to increase the efficiency to detect the pneumonia from the pictures. Statistical results show that the trained model was able to detect pneumonia by examining chest X-ray images