Dual-Scale Volume Priors with Wasserstein-Based Consistency for Semi-Supervised Medical Image Segmentation

2025-09-04

AI Summary

In semi-supervised medical image segmentation, feature learning suffers from insufficient guidance and underutilization of prior knowledge. Method: This paper proposes a framework integrating explicit and implicit dual-scale volumetric priors. Specifically: (1) it establishes image-level and dataset-level volumetric priors—estimating target region volume via regression and modeling population-level volume distribution, respectively; (2) it enforces cross-scale feature distribution alignment using a Wasserstein-distance-based consistency constraint; and (3) it incorporates spatial regularization derived from variational inference (Threshold Dynamics) to enhance structural coherence. Results: Experiments on ACDC, PROMISE12, and thigh muscle MR datasets demonstrate substantial improvements in segmentation accuracy and robustness under semi-supervised settings, particularly at extremely low labeling ratios (e.g., 1–5 labeled samples per dataset).

Technology Category

• Machine Learning: Semi-Supervised Learning
• Computer Vision: Segmentation
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Despite signi cant progress in semi-supervised medical image segmentation, most existing segmentation networks overlook e ective methodological guidance for feature extraction and important prior information from datasets. In this paper, we develop a semi-supervised medical image segmentation framework that e ectively integrates spatial regularization methods and volume priors. Speci cally, our approach integrates a strong explicit volume prior at the image scale and Threshold Dynamics spatial regularization, both derived from variational models, into the backbone segmentation network. The target region volumes for each unlabeled image are estimated by a regression network, which e ectively regularizes the backbone segmentation network through an image-scale Wasserstein distance constraint, ensuring that the class ratios in the segmentation results for each unlabeled image match those predicted by the regression network. Additionally, we design a dataset-scale Wasserstein distance loss function based on a weak implicit volume prior, which enforces that the volume distribution predicted for the unlabeled dataset is similar to that of labeled dataset. Experimental results on the 2017 ACDC dataset, PROMISE12 dataset, and thigh muscle MR image dataset show the superiority of the proposed method.

Problem

• Integrating spatial regularization and volume priors for semi-supervised segmentation
• Estimating target region volumes through regression network constraints
• Enforcing volume distribution consistency between labeled and unlabeled datasets

Innovation

• Dual-scale volume priors integration
• Wasserstein distance constraints application
• Spatial regularization with variational models