
This project presents an AI-powered chest X-ray classification system designed to assist in the early detection of pneumonia. We trained a ResNet101 deep learning model on chest X-ray images and achieved strong accuracy in distinguishing between Normal and Pneumonia cases. The trained model was converted into a TorchScript (.pt) format for portability and deployed using a Flask web application. Users can upload their chest X-ray images, and the system instantly provides a prediction, along with a user-friendly interface. The project demonstrates the potential of AI in healthcare by providing fast, reliable, and accessible diagnostic support. Future improvements could include multi-disease classification, integration with cloud platforms, and real-time hospital deployment. πΉ Problem Statement: Pneumonia is one of the leading causes of mortality worldwide, especially in children and the elderly. Traditional diagnosis through chest X-rays requires expert radiologists, which may not always be accessible in rural or under-resourced areas. Delayed diagnosis can lead to severe complications or death. There is a critical need for fast, accessible, and affordable tools to assist healthcare providers in detecting pneumonia. πΉ Solution: We developed an AI-based chest X-ray classification system using ResNet101, trained to distinguish between Normal and Pneumonia cases. The model was exported into a TorchScript (.pt) format for lightweight deployment and integrated into a Flask web application. Doctors or patients can simply upload a chest X-ray image, and the system instantly returns a prediction. πΉ Impact: Accessibility: Makes pneumonia detection available to clinics with limited radiology staff. Speed: Provides predictions in seconds, reducing diagnosis time. Scalability: Can be deployed globally via the web and reaching rural. Future Scope: Extendable to detect multiple lung diseases, integration with cloud-based hospital systems, and mobile app deployment.
21 Sep 2025