Scholarlink

<p>Accurate retinal vessel segmentation is crucial for ophthalmic image analysis, providing key structural information for diagnosis and treatment planning. However, existing methods struggle with multi-scale vessel variability, complex curvatures, and ambiguous boundaries. CNNs, Transformer, and Mamba-based approaches have shown promise, yet still struggle to maintain vascular continuity and accurately delineate fine vessel boundaries, especially in thin or tortuous regions, leading to structural discontinuities and edge ambiguity. To address these limitations, we propose a novel hybrid framework that synergistically integrates CNNs and Mamba for high-precision retinal vessel segmentation. Our approach introduces three key innovations: (1) The proposed High-Resolution Edge Fuse Network is a high-resolution preserving hybrid segmentation framework that enhances edge features to ensure accurate and robust vessel segmentation. (2) The Dynamic Snake Visual State Space block is designed to adaptively capture vessel curvature details and long-range dependencies. An improved eight-directional 2D Snake-Selective Scan mechanism and a dynamic weighting strategy enhance the perception of complex vascular topologies. (3) The MREF module enhances boundary precision through multi-scale edge feature aggregation, suppressing noise while emphasizing critical vessel structures across scales. Experiments on DRIVE, STARE, and CHASE_DB1 demonstrate the robust and effective performance of our method. Specifically, our approach attains Dice scores of 82.14%, 76.29%, and 80.46%; clDice scores of 82.40%, 80.30%, and 82.93%; and AUC values of 98.56%, 98.05%, and 98.78%, respectively. This work provides a robust method for clinical applications requiring accurate retinal vessel analysis. The code is available at <a href="https://github.com/frank-oy/HREFNet">https://github.com/frank-oy/HREFNet</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A novel hybrid approach for retinal vessel segmentation with dynamic long-range dependency and multi-scale retinal edge fusion enhancement

  • 期刊论文

  • 发表日期：

    2025年8月20日

查看全文
  • Yihao Ouyang,
  • Xunheng Kuang,
  • Mengjia Xiong,
  • Zhida Wang,
  • Yuanquan Wang

摘要

Accurate retinal vessel segmentation is crucial for ophthalmic image analysis, providing key structural information for diagnosis and treatment planning. However, existing methods struggle with multi-scale vessel variability, complex curvatures, and ambiguous boundaries. CNNs, Transformer, and Mamba-based approaches have shown promise, yet still struggle to maintain vascular continuity and accurately delineate fine vessel boundaries, especially in thin or tortuous regions, leading to structural discontinuities and edge ambiguity. To address these limitations, we propose a novel hybrid framework that synergistically integrates CNNs and Mamba for high-precision retinal vessel segmentation. Our approach introduces three key innovations: (1) The proposed High-Resolution Edge Fuse Network is a high-resolution preserving hybrid segmentation framework that enhances edge features to ensure accurate and robust vessel segmentation. (2) The Dynamic Snake Visual State Space block is designed to adaptively capture vessel curvature details and long-range dependencies. An improved eight-directional 2D Snake-Selective Scan mechanism and a dynamic weighting strategy enhance the perception of complex vascular topologies. (3) The MREF module enhances boundary precision through multi-scale edge feature aggregation, suppressing noise while emphasizing critical vessel structures across scales. Experiments on DRIVE, STARE, and CHASE_DB1 demonstrate the robust and effective performance of our method. Specifically, our approach attains Dice scores of 82.14%, 76.29%, and 80.46%; clDice scores of 82.40%, 80.30%, and 82.93%; and AUC values of 98.56%, 98.05%, and 98.78%, respectively. This work provides a robust method for clinical applications requiring accurate retinal vessel analysis. The code is available at https://github.com/frank-oy/HREFNet.