[论文解读] Multi-View Deformable Convolution Meets Visual Mamba for Coronary Artery Segmentation

Xiaochan Yuan, Pai Zeng|arXiv (Cornell University)|Mar 23, 2026

Retinal Imaging and Analysis被引用 0

一句话总结

本文提出 MDSVM-UNet，一种结合多向蛇形卷积（MDSConv）与残差视觉猛蛛（RVM）的两阶段冠状动脉分割模型，在 ImageCAS 上达到最先进的结果。

ABSTRACT

Accurate segmentation of coronary arteries from computed tomography angiography (CTA) images is of paramount clinical importance for the diagnosis and treatment planning of cardiovascular diseases. However, coronary artery segmentation remains challenging due to the inherent multi-branching and slender tubular morphology of the vasculature, compounded by severe class imbalance between foreground vessels and background tissue. Conventional convolutional neural network (CNN)-based approaches struggle to capture long-range dependencies among spatially distant vascular structures, while Vision Transformer (ViT)-based methods incur prohibitive computational overhead that hinders deployment in resource-constrained clinical settings. Motivated by the recent success of state space models (SSMs) in efficiently modeling long-range sequential dependencies with linear complexity, we propose MDSVM-UNet, a novel two-stage coronary artery segmentation framework that synergistically integrates multidirectional snake convolution (MDSConv) with residual visual Mamba (RVM). In the encoding stage, we introduce MDSConv, a deformable convolution module that learns adaptive offsets along three orthogonal anatomical planes -- sagittal, coronal, and axial -- thereby enabling comprehensive multi-view feature fusion that faithfully captures the elongated and tortuous geometry of coronary vessels. In the decoding stage, we design an RVM-based upsampling decoder block that leverages selective state space mechanisms to model inter-slice long-range dependencies while preserving linear computational complexity. Furthermore, we propose a progressive two-stage segmentation strategy: the first stage performs coarse whole-image segmentation to guide intelligent block extraction, while the second stage conducts fine-grained block-level segmentation to recover vascular details and suppress false positives..

研究动机与目标

• 促成从 CTA 自动实现准确的冠状动脉分割，以辅助诊断与治疗计划。
• 解决管状、细长的血管形态以及 foreground-background 的严重不平衡。
• 提出一个两阶段的粗到细框架，在全局上下文与局部血管细节之间实现平衡。

提出的方法

• 引入 MDSConv，在解剖三个正交平面（矢状面、冠状面、横断面）上进行可变形卷积，并融合多视角特征以适应管状结构。
• 使用残差视觉猛蛛解码器（RVM）对跨切片的长程依赖进行建模，线性复杂度。
• 采用两阶段渐进分割：先进行粗略全图分割以引导块级提取，再进行精细的块级分割。
• 采用 UNet++ 风格的密集跳跃连接，以实现丰富的多尺度特征传播。
• 以 Dice 损失进行训练，以应对血管与背景之间的严重类别不平衡。

实验结果

研究问题

• RQ1 多方向的可变形卷积是否能比标准 CNN 更好地捕捉延展且迂回的冠状血管？
• RQ2 将残差视觉猛蛛解码器整合是否能在线性复杂度下提升长程依赖建模与分割质量？
• RQ3 两阶段的粗到细策略是否比单阶段方法在血管连续性和减少假阳性方面有改进？
• RQ4 在 ImageCAS 基准上，MDSVM-UNet 相对于最先进方法在 DSC、HD、AHD 方面的表现如何？

主要发现

• MDSVM-UNet 在单阶段达到 0.686 的 DSC，在两阶段配置（阶段 2 使用 Dice 损失）达到 0.8365 的 DSC。
• 两阶段的 MDSVM-UNet 在 DSC 提升 5.41%、HD 降幅 8.5456、AHD 降幅 0.8093，相较 ImageCAS 基线。
• 阶段 1（单阶段）Dice 损失时的 DSC 为 0.6860，HD 为 27.8430；AHD 为 0.9023（表 1）。
• 两阶段的 MDSVM-UNet 使用 Dice 损失时达到 DSC 0.8365、HD 27.8430、AHD 0.9023（表 2）。
• 模型参数量为 2670 万，与基于 Transformer 的方法相比具有竞争力的效率。