Path-Weighted Integrated Gradients for Interpretable Dementia Classification

2025-09-22

AI Summary

Integrated Gradients (IG) suffer from uniform sensitivity across the baseline-to-input path, vulnerability to noise, and inability to capture path-dependent features in attribution. To address these limitations, we propose Path-Weighted Integrated Gradients (PWIG), which augments the IG framework with a learnable or customizable weighting function to assign differential weights to path segments, thereby enabling selective attribution to salient regions. Evaluated on dementia classification using the OASIS-1 MRI dataset, PWIG produces attribution maps that significantly improve clinical interpretability and cross-sample stability—precisely highlighting disease-relevant brain regions (e.g., hippocampus and prefrontal cortex) while suppressing spurious noise responses. This work constitutes the first systematic integration of path-weighting into IG to enhance modeling of path-dependent attribution behavior. It establishes a novel, robust paradigm for eXplainable AI (XAI) in medical imaging, advancing both fidelity and reliability of model explanations.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Humans and AI: Explainable AI (XAI) for Human Understanding
• Machine Learning: Transparent, Interpretable, Explainable ML

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Integrated Gradients (IG) is a widely used attribution method in explainable artificial intelligence (XAI). In this paper, we introduce Path-Weighted Integrated Gradients (PWIG), a generalization of IG that incorporates a customizable weighting function into the attribution integral. This modification allows for targeted emphasis along different segments of the path between a baseline and the input, enabling improved interpretability, noise mitigation, and the detection of path-dependent feature relevance. We establish its theoretical properties and illustrate its utility through experiments on a dementia classification task using the OASIS-1 MRI dataset. Attribution maps generated by PWIG highlight clinically meaningful brain regions associated with various stages of dementia, providing users with sharp and stable explanations. The results suggest that PWIG offers a flexible and theoretically grounded approach for enhancing attribution quality in complex predictive models.

Problem

• Generalizing Integrated Gradients with customizable path weighting functions
• Improving interpretability and noise mitigation in dementia classification
• Detecting path-dependent feature relevance for clinical MRI analysis

Innovation

• Generalizes Integrated Gradients with a customizable weighting function
• Enables targeted emphasis along the feature attribution path
• Produces sharp, stable attribution maps for clinical interpretability