Semantic-aware Random Convolution and Source Matching for Domain Generalization in Medical Image Segmentation

2025-12-01

AI Summary

This paper addresses single-source domain generalization (SSDG) for medical image segmentation—i.e., training a model on a single source domain (e.g., CT) and achieving robust cross-modal (e.g., to MR), multi-center, and multi-cardiac-phase segmentation without access to target-domain data or fine-tuning. We propose Semantic-Aware Random Convolution (SARConv), which applies anatomy-guided, label-aware augmentation to source images, and a source-domain matching intensity mapping strategy that adaptively calibrates target-domain intensity distributions during inference. Integrated into mainstream segmentation architectures, these components jointly mitigate semantic and distributional shifts between domains. Evaluated on multiple cross-modal benchmarks, our method achieves state-of-the-art performance, with segmentation accuracy in certain scenarios approaching that of in-domain supervised baselines—establishing a new benchmark for SSDG in medical image segmentation.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

We tackle the challenging problem of single-source domain generalization (DG) for medical image segmentation. To this end, we aim for training a network on one domain (e.g., CT) and directly apply it to a different domain (e.g., MR) without adapting the model and without requiring images or annotations from the new domain during training. We propose a novel method for promoting DG when training deep segmentation networks, which we call SRCSM. During training, our method diversifies the source domain through semantic-aware random convolution, where different regions of a source image are augmented differently, based on their annotation labels. At test-time, we complement the randomization of the training domain via mapping the intensity of target domain images, making them similar to source domain data. We perform a comprehensive evaluation on a variety of cross-modality and cross-center generalization settings for abdominal, whole-heart and prostate segmentation, where we outperform previous DG techniques in a vast majority of experiments. Additionally, we also investigate our method when training on whole-heart CT or MR data and testing on the diastolic and systolic phase of cine MR data captured with different scanner hardware, where we make a step towards closing the domain gap in this even more challenging setting. Overall, our evaluation shows that SRCSM can be considered a new state-of-the-art in DG for medical image segmentation and, moreover, even achieves a segmentation performance that matches the performance of the in-domain baseline in several settings.

Problem

• Single-source domain generalization for medical image segmentation
• Training on one domain and applying to another without adaptation
• Closing domain gaps across modalities and scanner hardware

Innovation

• Semantic-aware random convolution diversifies source domain training
• Intensity mapping aligns target domain images with source data
• SRCSM outperforms prior domain generalization techniques in segmentation