Single Domain Generalization for Alzheimer's Detection from 3D MRIs with Pseudo-Morphological Augmentations and Contrastive Learning

2025-05-28

AI Summary

Alzheimer’s disease (AD) diagnosis via single-domain generalization (SDG) suffers from poor cross-center transferability, severely hindered by inter-scanner protocol variations and class imbalance. Method: We propose a supervised contrastive learning framework incorporating learnable pseudo-morphological augmentation. Its core is the first differentiable, anatomically plausible, and class-specific 3D pseudo-morphological module, explicitly encoding regional brain shape priors; combined with a morphology-aware supervised contrastive loss, it learns robust, generalizable representations from single-center MRI data alone. Training requires no target-domain data or additional annotations. Contribution/Results: Evaluated on three heterogeneous multi-center MRI datasets, our method significantly improves cross-center AD classification performance. It effectively mitigates generalization degradation induced by protocol heterogeneity and label skew, delivering high clinical robustness for real-world deployment.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Adversarial Attacks & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Although Alzheimer's disease detection via MRIs has advanced significantly thanks to contemporary deep learning models, challenges such as class imbalance, protocol variations, and limited dataset diversity often hinder their generalization capacity. To address this issue, this article focuses on the single domain generalization setting, where given the data of one domain, a model is designed and developed with maximal performance w.r.t. an unseen domain of distinct distribution. Since brain morphology is known to play a crucial role in Alzheimer's diagnosis, we propose the use of learnable pseudo-morphological modules aimed at producing shape-aware, anatomically meaningful class-specific augmentations in combination with a supervised contrastive learning module to extract robust class-specific representations. Experiments conducted across three datasets show improved performance and generalization capacity, especially under class imbalance and imaging protocol variations. The source code will be made available upon acceptance at https://github.com/zobia111/SDG-Alzheimer.

Problem

• Improving Alzheimer's detection generalization from single-domain MRI data
• Addressing class imbalance and protocol variations in MRI datasets
• Enhancing shape-aware feature learning via pseudo-morphological augmentations

Innovation

• Pseudo-morphological augmentations for shape-aware MRI data
• Supervised contrastive learning for robust representations
• Single domain generalization for unseen data distribution