HarmonySeg: Tubular Structure Segmentation with Deep-Shallow Feature Fusion and Growth-Suppression Balanced Loss

2025-04-10

AI Summary

Accurate segmentation of tubular structures (e.g., vessels, airways) in medical images remains challenging due to multi-scale variations, complex topologies, and sparse annotations. To address these issues, we propose a robust segmentation framework featuring: (1) a deep-shallow feature fusion decoder to enhance multi-scale representation; (2) a vesselness-guided auxiliary alignment module that precisely couples shallow-level localization cues with deep-level semantic features; and (3) a growth-suppression balanced loss function jointly modeling topological priors and robustness to annotation noise. The framework further incorporates flexible-receptive-field convolutions and a context- and shape-aware loss for optimization. Evaluated on four public benchmarks, our method consistently outperforms state-of-the-art approaches in both 2D and 3D settings. Moreover, it demonstrates strong generalization capability on a private clinical dataset, confirming its practical applicability across diverse imaging domains and annotation regimes.

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: 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 neural networks and deep learning approaches for Web-related graphs

Abstract

Accurate segmentation of tubular structures in medical images, such as vessels and airway trees, is crucial for computer-aided diagnosis, radiotherapy, and surgical planning. However, significant challenges exist in algorithm design when faced with diverse sizes, complex topologies, and (often) incomplete data annotation of these structures. We address these difficulties by proposing a new tubular structure segmentation framework named HarmonySeg. First, we design a deep-to-shallow decoder network featuring flexible convolution blocks with varying receptive fields, which enables the model to effectively adapt to tubular structures of different scales. Second, to highlight potential anatomical regions and improve the recall of small tubular structures, we incorporate vesselness maps as auxiliary information. These maps are aligned with image features through a shallow-and-deep fusion module, which simultaneously eliminates unreasonable candidates to maintain high precision. Finally, we introduce a topology-preserving loss function that leverages contextual and shape priors to balance the growth and suppression of tubular structures, which also allows the model to handle low-quality and incomplete annotations. Extensive quantitative experiments are conducted on four public datasets. The results show that our model can accurately segment 2D and 3D tubular structures and outperform existing state-of-the-art methods. External validation on a private dataset also demonstrates good generalizability.

Problem

• Segments tubular structures in medical images accurately
• Handles diverse sizes and complex topologies effectively
• Improves performance with incomplete data annotations

Innovation

• Deep-to-shallow decoder with flexible convolution blocks
• Vesselness maps fused via shallow-deep module
• Topology-preserving loss balancing growth-suppression