RARE-UNet: Resolution-Aligned Routing Entry for Adaptive Medical Image Segmentation

2025-07-21

AI Summary

Medical image segmentation models suffer significant performance degradation under input resolution variations. To address this, we propose a resolution-adaptive multi-scale segmentation architecture that enables dynamic alignment between cross-resolution features and full-resolution representations—first of its kind. Built upon the U-Net framework, our method introduces a resolution-aware routing mechanism jointly optimized with multi-scale encoding modules to enable dynamic path selection, and incorporates a consistency-driven training strategy to enhance robustness in multi-scale feature fusion. Evaluated on hippocampus and tumor segmentation tasks, our model achieves mean Dice scores of 0.84 and 0.65, respectively—substantially outperforming both U-Net and nnU-Net. Notably, it maintains high accuracy at low resolutions while achieving significantly accelerated inference. Key contributions include: (1) a resolution-aware dynamic routing mechanism; (2) effective cross-scale feature alignment; and (3) a consistency-driven adaptive training paradigm.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Accurate segmentation is crucial for clinical applications, but existing models often assume fixed, high-resolution inputs and degrade significantly when faced with lower-resolution data in real-world scenarios. To address this limitation, we propose RARE-UNet, a resolution-aware multi-scale segmentation architecture that dynamically adapts its inference path to the spatial resolution of the input. Central to our design are multi-scale blocks integrated at multiple encoder depths, a resolution-aware routing mechanism, and consistency-driven training that aligns multi-resolution features with full-resolution representations. We evaluate RARE-UNet on two benchmark brain imaging tasks for hippocampus and tumor segmentation. Compared to standard UNet, its multi-resolution augmented variant, and nnUNet, our model achieves the highest average Dice scores of 0.84 and 0.65 across resolution, while maintaining consistent performance and significantly reduced inference time at lower resolutions. These results highlight the effectiveness and scalability of our architecture in achieving resolution-robust segmentation. The codes are available at: https://github.com/simonsejse/RARE-UNet.

Problem

• Adapts segmentation to varying input resolutions dynamically
• Improves accuracy across multi-resolution medical imaging data
• Reduces inference time while maintaining consistent performance

Innovation

• Resolution-aware multi-scale segmentation architecture
• Dynamic inference path adaptation
• Consistency-driven multi-resolution feature alignment