AI-Powered Chest X-Ray Diagnostic Assistant with Explainable AI

This project focuses on building an AI-powered diagnostic assistant for chest X-ray interpretation, designed to support radiologists and clinicians in detecting pulmonary abnormalities with greater speed and confidence.

Technical Approach: The system leverages DenseNet-121 architecture for multi-label classification of chest pathologies, trained on large-scale datasets to identify conditions including pneumonia, cardiomegaly, pleural effusion, and more.

Explainability at the Core: Unlike black-box models, this tool integrates Grad-CAM (Gradient-weighted Class Activation Mapping) to generate visual heatmaps highlighting the regions influencing each prediction — enabling clinicians to understand why the AI flagged an area, not just what it flagged.

Collaboration Opportunities:

• Clinical validation and real-world testing

• Dataset contribution and annotation

• UI/UX development for clinical workflows

• Research co-authorship and publication

• Expansion to other imaging modalities (CT, MRI)

Looking for: Medical professionals, radiologists, AI/ML researchers, healthcare institutions, and students passionate about bridging AI and clinical practice.

Current Stage: Functional prototype with deployed demo. Seeking partners for validation, expansion, and potential publication.