HybridMamba: A Dual-domain Mamba for 3D Medical Image Segmentation

2025-09-18

AI Summary

Existing Mamba-based models for 3D medical image segmentation overemphasize global modeling at the expense of local structural fidelity, leading to blurred boundaries and distorted anatomical regions. Method: We propose a dual-domain Mamba architecture featuring (1) a local-adaptive and axial-traversal dual-path feature scanning mechanism to enhance multi-scale local perception, and (2) a gated spatial–frequency analysis module that dynamically fuses spatial details with global context under frequency-domain guidance. Contribution/Results: The architecture preserves Mamba’s computational efficiency and long-range modeling capability while significantly improving local structure preservation. Evaluated on multi-center CT and diverse MRI datasets, our method achieves state-of-the-art segmentation accuracy and cross-device robustness with fewer parameters, establishing a new paradigm for lightweight and efficient 3D medical image segmentation.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Deep Neural Architectures and Foundation Models
• 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
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

In the domain of 3D biomedical image segmentation, Mamba exhibits the superior performance for it addresses the limitations in modeling long-range dependencies inherent to CNNs and mitigates the abundant computational overhead associated with Transformer-based frameworks when processing high-resolution medical volumes. However, attaching undue importance to global context modeling may inadvertently compromise critical local structural information, thus leading to boundary ambiguity and regional distortion in segmentation outputs. Therefore, we propose the HybridMamba, an architecture employing dual complementary mechanisms: 1) a feature scanning strategy that progressively integrates representations both axial-traversal and local-adaptive pathways to harmonize the relationship between local and global representations, and 2) a gated module combining spatial-frequency analysis for comprehensive contextual modeling. Besides, we collect a multi-center CT dataset related to lung cancer. Experiments on MRI and CT datasets demonstrate that HybridMamba significantly outperforms the state-of-the-art methods in 3D medical image segmentation.

Problem

• Addresses long-range dependency limitations in 3D medical segmentation
• Mitigates computational overhead of Transformer-based 3D frameworks
• Resolves boundary ambiguity from excessive global context modeling

Innovation

• Dual complementary mechanisms for segmentation
• Feature scanning strategy integrating axial and local pathways
• Gated spatial-frequency module for contextual modeling