Utilising Explainability techniques in Medical Imaging

MICCAI 2024 MEC Submission13 Authors

Keywords: explainable ai, medical imaging

Abstract:

The goal of this tutorial is to implement explainable AI (XAI) techniques that are crucial for understanding and interpreting the decisions made by complex deep learning models, particularly in medical imaging.

In medical imaging, the transparency of AI models is critical for building trust with healthcare professionals and ensuring the safe and effective use of AI-driven diagnostics. This tutorial will guide you through implementing two fundamental XAI techniques: Gradient-weighted Class Activation Mapping (Grad-CAM) and SHapley Additive exPlanations (SHAP). These methods provide visual and feature-level insights into model predictions, helping to demystify the "black box" nature of AI models.

Website: https://colab.research.google.com/drive/1aSOZiqr7ZVq51XxazvK8ikuKy9UmJHa2?usp=sharing

Submission Number: 13