Homeomorphism Prior for False Positive and Negative Problem in Medical Image Dense Contrastive Representation Learning

2025-02-07

AI Summary

Dense contrastive representation learning (DCRL) for medical images faces challenges from unreliable pixel-wise correspondences due to anatomical deformations and pervasive false-positive/negative sample selection. To address this, we propose the first homoeomorphic prior-integrated DCRL framework—Deformable Homeomorphic Learning (DHL) coupled with Geometric-Semantic Similarity (GSS). DHL enforces topologically preserved deformable mappings for robust pixel alignment, while GSS jointly leverages geometric consistency and semantic similarity to soften negative sampling and enhance positive pair reliability. Our method incorporates topological constraint optimization, dense contrastive loss, and gradient-driven implicit negative sampling. Evaluated across seven diverse medical imaging datasets, our approach achieves state-of-the-art performance in both segmentation and registration tasks, demonstrating superior generalizability and robustness over existing methods.

Technology Category

• Computer Vision: Medical and Biological Imaging
• Machine Learning: Representation Learning
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

Dense contrastive representation learning (DCRL) has greatly improved the learning efficiency for image-dense prediction tasks, showing its great potential to reduce the large costs of medical image collection and dense annotation. However, the properties of medical images make unreliable correspondence discovery, bringing an open problem of large-scale false positive and negative (FP&N) pairs in DCRL. In this paper, we propose GEoMetric vIsual deNse sImilarity (GEMINI) learning which embeds the homeomorphism prior to DCRL and enables a reliable correspondence discovery for effective dense contrast. We propose a deformable homeomorphism learning (DHL) which models the homeomorphism of medical images and learns to estimate a deformable mapping to predict the pixels' correspondence under topological preservation. It effectively reduces the searching space of pairing and drives an implicit and soft learning of negative pairs via a gradient. We also propose a geometric semantic similarity (GSS) which extracts semantic information in features to measure the alignment degree for the correspondence learning. It will promote the learning efficiency and performance of deformation, constructing positive pairs reliably. We implement two practical variants on two typical representation learning tasks in our experiments. Our promising results on seven datasets which outperform the existing methods show our great superiority. We will release our code on a companion link: https://github.com/YutingHe-list/GEMINI.

Problem

• Reducing false positives and negatives
• Improving dense contrastive representation learning
• Enhancing medical image correspondence discovery

Innovation

• Embeds homeomorphism prior
• Models deformable homeomorphism learning
• Extracts geometric semantic similarity