Evaluating and Selecting Optimal CNN Architectures for Accurate Pneumonia Detection in Chest X-Rays

Pneumonia is the lungs' alveoli filling with fluid, and it mainly affects children below 5 years and adults above 65. Results: We demonstrate our approach using different configurations of convolutional neural networks (CNNs) on a chest X-ray binary classification task detecting pneumonia cases. The focus here is more on performance evaluation among various simple CNN architectures to find the one that gives least loss and highest accuracy. The ultimate aim is to enable the widespread adoption of a strong tool for the diagnosis of viral, bacterial and fungal pneumonia as well as community-acquired pneumonia based on chest X-rays only by clinicians.