Causally Guided Gaussian Perturbations for Out-Of-Distribution Generalization in Medical Imaging

2025-09-30

AI Summary

Medical imaging faces out-of-distribution (OOD) generalization challenges characterized by subtle yet pervasive distribution shifts; existing approaches rely on complex generative models or adversarial training, often neglecting underlying causal mechanisms. To address this, we propose a lightweight, causality-guided Gaussian perturbation framework: a vision Transformer generates soft causal masks to enable spatially adaptive noise injection—reducing perturbation intensity in causally relevant regions while amplifying it in spurious ones—thereby suppressing reliance on non-causal features. Our method requires no auxiliary domain labels or architectural modifications, significantly enhancing model robustness and interpretability. Evaluated on the Camelyon17 benchmark, it surpasses state-of-the-art OOD generalization methods with substantially fewer parameters, demonstrating the efficacy and practicality of causality-driven input perturbation for biomedical image analysis.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Data transparency and provenance
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Out-of-distribution (OOD) generalization remains a central challenge in deploying deep learning models to real-world scenarios, particularly in domains such as biomedical images, where distribution shifts are both subtle and pervasive. While existing methods often pursue domain invariance through complex generative models or adversarial training, these approaches may overlook the underlying causal mechanisms of generalization.In this work, we propose Causally-Guided Gaussian Perturbations (CGP)-a lightweight framework that enhances OOD generalization by injecting spatially varying noise into input images, guided by soft causal masks derived from Vision Transformers. By applying stronger perturbations to background regions and weaker ones to foreground areas, CGP encourages the model to rely on causally relevant features rather than spurious correlations.Experimental results on the challenging WILDS benchmark Camelyon17 demonstrate consistent performance gains over state-of-the-art OOD baselines, highlighting the potential of causal perturbation as a tool for reliable and interpretable generalization.

Problem

• Improves out-of-distribution generalization for medical imaging
• Reduces reliance on spurious correlations using causal guidance
• Applies spatially varying perturbations via Vision Transformer masks

Innovation

• Causally-guided Gaussian perturbations enhance generalization
• Spatially varying noise guided by Vision Transformer masks
• Stronger background perturbations focus on causal features