CSWinUNETR: Segmentation of Thin Anatomical Structures in Medical Images

2026-06-18

AI Summary

This work addresses the challenging segmentation of thin, curvilinear anatomical structures—such as retinal vessels, cerebral vasculature, and facial wrinkles—in medical images, which is hindered by low contrast, discontinuities, and severe class imbalance. To this end, the authors propose CSWinUNETR, a general-purpose backbone network that integrates cross-shaped strip self-attention with a cyclic shift mechanism to effectively model long-range axial contextual dependencies. A detail-enhanced multi-scale self-attention module is designed to fuse features across resolutions, while a novel sparse-controlled dynamic snake convolution is introduced for the first time to reconstruct dense curve kernels from sparse control points, thereby conforming precisely to complex geometric shapes. Without requiring task-specific post-processing or topology-aware losses, the method recovers fine branches and preserves structural continuity. Extensive experiments on four cross-modality medical benchmarks demonstrate that CSWinUNETR significantly outperforms state-of-the-art models, achieving superior completeness and accuracy in fine structure segmentation.

Abstract

Accurate segmentation of thin, tortuous anatomical structures, such as retinal vessels, cerebral vasculature, and facial wrinkles, remains challenging due to low contrast, frequent discontinuities, and severe class imbalance. Although recent convolutional and Transformer-based models have improved performance, they often yield fragmented predictions and fail to recover fine branches. We propose CSWinUNETR, a general-purpose backbone for 2D and 3D thin-structure segmentation. It employs cross-shaped stripe self-attention to model long-range principal-axis context and incorporates cyclic shifts to enhance information exchange across stripes. To better preserve fine-grained details, we further introduce a detail-enhanced multi-scale self-attention module that aggregates contextual features from multi-resolution representations. In addition, we propose sparse-control dynamic snake convolution, which reconstructs reliable dense curvilinear kernels from sparsely predicted control points to better follow tortuous geometry. Extensive experiments on four benchmarks across ophthalmology, neurovascular imaging, and dermatology demonstrate that CSWinUNETR consistently outperforms state-of-the-art methods without task-specific post-processing or topology-aware losses. The code is available at https://github.com/labhai/CSWinUNETR.

Problem

• thin anatomical structures
• medical image segmentation
• low contrast
• class imbalance
• tortuous geometry

Innovation

• cross-shaped stripe self-attention
• detail-enhanced multi-scale self-attention
• sparse-control dynamic snake convolution
• thin-structure segmentation
• medical image analysis