Multi Disease Diagnostic Analysis for Chest X-Ray Images with Explainable AI

Abstract

Integrating Explainable Artificial Intelligence (XAI) in the Multi Disease Dia- gnostic Model for Chest X-ray Images enables clearer, more interpretable diagnoses for stakeholders. By providing insights into model reasoning, XAI enhances trust and accountability, improving diagnostic transparency and accuracy, which is critical for healthcare providers and patients alike. In this project, various XAI tools such as LIME, GRAD-CAM were applied to improve reliability and interpretability in multi-disease diagnosis using chest X-ray images. An open-source dataset comprising chest X-ray images and associated metadata was used, and preprocessing techniques such as resizing, normalization, and data augmentation were applied to ensure data quality and relevance. Deep learning models were implemented, leveraging each model’s strengths to improve diagnostic accuracy. The XAI methods LIME and GRAD-CAM were combined to improve model transparency and offer comprehensible findings about how decisions are made of the model. The interpretability aids various stakeholders, such as healthcare professionals, patients and data analysts, in understanding the reasoning behind model predictions, fostering trust and enabling more precise, data driven decisions in multi-disease diagnostic scenarios.