The Use of Chest Radiographs and Machine Learning Model for the Rapid Detection of Pneumonitis in Pediatric

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computer Science

Background:

• Pneumonia is a leading global cause of death, particularly affecting vulnerable populations like children and the elderly.
• Chest radiography remains a primary diagnostic tool for pulmonary infections due to its cost-effectiveness and speed.
• Existing diagnostic methods for pneumonia have limitations, necessitating advancements in detection technology.

Purpose of the Study:

• To develop and evaluate a convolutional neural network (CNN) model for classifying chest X-rays into normal, bacterial pneumonia, and viral pneumonia categories.
• To assess the impact of data augmentation on the CNN model's performance and diagnostic accuracy.
• To create a user-friendly web application for AI-assisted pneumonia detection.

Main Methods:

• A CNN model was trained on a Kaggle dataset of chest X-ray images in JPEG format.
• Data augmentation techniques were employed to enhance the model's robustness and predictive capabilities.
• A web application was developed using NextJS and hosted on AWS to deploy the trained model for real-time image analysis.

Main Results:

• The CNN model achieved a classification accuracy of 78% for detecting pneumonia.
• Data augmentation resulted in a slight improvement in the model's precision compared to the original model.
• The developed web application successfully processes X-ray images and provides diagnostic predictions.

Conclusions:

• Deep learning models, such as the developed CNN, show promise in assisting healthcare professionals with the early detection of pneumonia.
• Further optimization, including increasing training epochs, could potentially enhance the model's precision.
• AI-powered diagnostic tools can augment clinical decision-making for pulmonary infections.