Unleashing the Power of Intermediate Domains for Mixed Domain Semi-Supervised Medical Image Segmentation

2025-05-30

AI Summary

Addressing the dual challenges of **label scarcity** and **multi-source domain shift** in medical image segmentation, this paper proposes a novel hybrid-domain semi-supervised segmentation paradigm (MiDSS). Under the realistic setting where only a few labeled samples are available in a single source domain and unlabeled data reside across multiple target domains, MiDSS constructs a unified intermediate domain to bridge heterogeneous domain distributions. Methodologically, it introduces the first intermediate-domain-driven framework, featuring symmetrically guided pseudo-label fusion, training-aware random-magnitude MixUp, and reliability-based differential intermediate sample generation. It further integrates uncertainty-aware consistency regularization (UCP), pseudo-label quality filtering, and reweighting strategies. Extensive experiments on four public benchmarks demonstrate substantial improvements: e.g., a 12.94% Dice score gain for prostate segmentation. The code is publicly available.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Semi-Supervised Learning
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Security and Privacy: Data transparency and provenance

Abstract

Both limited annotation and domain shift are prevalent challenges in medical image segmentation. Traditional semi-supervised segmentation and unsupervised domain adaptation methods address one of these issues separately. However, the coexistence of limited annotation and domain shift is quite common, which motivates us to introduce a novel and challenging scenario: Mixed Domain Semi-supervised medical image Segmentation (MiDSS), where limited labeled data from a single domain and a large amount of unlabeled data from multiple domains. To tackle this issue, we propose the UST-RUN framework, which fully leverages intermediate domain information to facilitate knowledge transfer. We employ Unified Copy-paste (UCP) to construct intermediate domains, and propose a Symmetric GuiDance training strategy (SymGD) to supervise unlabeled data by merging pseudo-labels from intermediate samples. Subsequently, we introduce a Training Process aware Random Amplitude MixUp (TP-RAM) to progressively incorporate style-transition components into intermediate samples. To generate more diverse intermediate samples, we further select reliable samples with high-quality pseudo-labels, which are then mixed with other unlabeled data. Additionally, we generate sophisticated intermediate samples with high-quality pseudo-labels for unreliable samples, ensuring effective knowledge transfer for them. Extensive experiments on four public datasets demonstrate the superiority of UST-RUN. Notably, UST-RUN achieves a 12.94% improvement in Dice score on the Prostate dataset. Our code is available at https://github.com/MQinghe/UST-RUN

Problem

• Addresses limited annotation and domain shift in medical image segmentation
• Proposes MiDSS scenario with labeled and unlabeled multi-domain data
• Introduces UST-RUN framework for effective knowledge transfer

Innovation

• Unified Copy-paste constructs intermediate domains
• Symmetric GuiDance merges pseudo-labels supervision
• Training Process aware Random Amplitude MixUp