Pneumonia Detection using Chest X-Ray Images: A Convolutional Neural Network with Skip Connection Approach

Pneumonia is a lung infection, that causes breathing difficulties and significant risks to young children, older persons, and individuals with weak immune systems. Diagnosing pneumonia typically involves examining chest X-rays to identify abnormal white patches over the lung area that indicate infection. This study addresses the diagnosis of pneumonia using artificial intelligence techniques, thereby reducing the need for extensive human intervention and saving time in pneumonia diagnosis. To achieve this, we developed a Convolutional Neural Network (CNN) model with skip connection blocks. These skip connections help the model retain critical information and accelerate convergence thus eliminating the vanishing gradient problem. The model was trained on 5,863 chest radiograph images, achieving outstanding accuracy, precision, and recall scores of 96.76%, 97.56%, and 98.01%, respectively. This high performance ensures accurate pneumonia identification, in the areas where precise diagnosis is crucial, such as in healthcare facilities. The study's findings emphasize the effectiveness of residual blocks in improving CNN performance for medical image analysis. To emphasise the effectiveness of our model we have also implemented comparative study with other established models including InceptionV3, ResNet50 and VGG-16. Proposed model demonstrated superiority in convergence time and model parameter efficiency. By reducing training iterations, our model expedites the diagnostic process and underscores the potential of AI in rapid medical diagnostics. This pioneering research establishes a strong foundation for future studies aimed at leveraging AI in diagnosing a variety of medical conditions using X-ray imaging. It has the potential to revolutionize medical diagnostics through advanced computational techniques.