Multi-Class Lung Disease Detection Using Resnet50 and Transfer Learning

Lung diseases, including Viral Pneumonia, Bacterial Pneumonia, COVID-19, Tuberculosis, and other common conditions, pose challenges in medical diagnostics. This study introduces an innovative method that merges deep learning techniques for classifying lung diseases with an AI-powered symptom checker to improve diagnostic precision and patient involvement. The proposed constructed a CNN ResNet50 architecture to categorize chest X-ray images into five distinct classes. The model underwent training and validation using an extensive dataset of labelled X-ray images, resulting in accuracy of 86% and strong performance metrics. To augment image-based diagnostic process, an incorporated generative AI model that offers users comprehensive information regarding symptoms and possible conditions based on their input. The system was implemented through user-friendly web application using Streamlit, facilitating real-time interaction and feedback. This integrated strategy aims to assist healthcare professionals in making better-informed decisions while providing patients with valuable health insights.