TopoCL: Topological Contrastive Learning for Medical Imaging

2026-03-15

AI Summary

This work proposes TopoCL, a novel contrastive learning framework that addresses the limitation of existing methods in medical image analysis, which typically rely solely on visual appearance while neglecting critical topological structures such as connectivity, cavities, and boundary configurations. TopoCL is the first to integrate topological persistence diagrams into contrastive learning by introducing topology-aware data augmentation, a hierarchical topological encoder, and an adaptive mixture-of-experts module to dynamically fuse visual and topological representations. The framework is compatible with mainstream contrastive learning architectures and demonstrates consistent improvements across five algorithms and five medical image classification datasets, achieving an average 3.26% gain in linear probing accuracy (p < 0.01) while preserving essential topological properties and significantly enhancing representational capacity.

Technology Category

• Computer Vision: Representation Learning for Vision
• Machine Learning: Representation Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Contrastive learning (CL) has become a powerful approach for learning representations from unlabeled images. However, existing CL methods focus predominantly on visual appearance features while neglecting topological characteristics (e.g., connectivity patterns, boundary configurations, cavity formations) that provide valuable cues for medical image analysis. To address this limitation, we propose a new topological CL framework (TopoCL) that explicitly exploits topological structures during contrastive learning for medical imaging. Specifically, we first introduce topology-aware augmentations that control topological perturbations using a relative bottleneck distance between persistence diagrams, preserving medically relevant topological properties while enabling controlled structural variations. We then design a Hierarchical Topology Encoder that captures topological features through self-attention and cross-attention mechanisms. Finally, we develop an adaptive mixture-of-experts (MoE) module to dynamically integrate visual and topological representations. TopoCL can be seamlessly integrated with existing CL methods. We evaluate TopoCL on five representative CL methods (SimCLR, MoCo-v3, BYOL, DINO, and Barlow Twins) and five diverse medical image classification datasets. The experimental results show that TopoCL achieves consistent improvements: an average gain of +3.26% in linear probe classification accuracy with strong statistical significance, verifying its effectiveness.

Problem

• contrastive learning
• topological features
• medical imaging
• representation learning
• topological structures

Innovation

• Topological Contrastive Learning
• Persistence Diagrams
• Topology-aware Augmentation
• Hierarchical Topology Encoder
• Mixture-of-Experts