DB-KAUNet: An Adaptive Dual Branch Kolmogorov-Arnold UNet for Retinal Vessel Segmentation

TL;DR

DB-KAUNet tackles retinal vessel segmentation by integrating Kolmogorov–Arnold networks into a dual-branch CNN-Transformer architecture. The Heterogeneous Dual-Branch Encoder (HDBE) interleaves standard and KAN modules, with Cross-Branch Channel Interaction and geometry-aware Spatial Feature Enhancement (SFE-GAF) to fuse features efficiently. Key innovations include KANConv/KAT blocks, LDConv-based SFE-GAF, and an interleaved decoder, achieving state-of-the-art performance on DRIVE, STARE, and CHASE_DB1 with strong robustness to noise and artifacts. The approach delivers high segmentation quality while maintaining favorable computational efficiency, demonstrating practical potential for clinical retinal analysis.

Abstract

Accurate segmentation of retinal vessels is crucial for the clinical diagnosis of numerous ophthalmic and systemic diseases. However, traditional Convolutional Neural Network (CNN) methods exhibit inherent limitations, struggling to capture long-range dependencies and complex nonlinear relationships. To address the above limitations, an Adaptive Dual Branch Kolmogorov-Arnold UNet (DB-KAUNet) is proposed for retinal vessel segmentation. In DB-KAUNet, we design a Heterogeneous Dual-Branch Encoder (HDBE) that features parallel CNN and Transformer pathways. The HDBE strategically interleaves standard CNN and Transformer blocks with novel KANConv and KAT blocks, enabling the model to form a comprehensive feature representation. To optimize feature processing, we integrate several critical components into the HDBE. First, a Cross-Branch Channel Interaction (CCI) module is embedded to facilitate efficient interaction of channel features between the parallel pathways. Second, an attention-based Spatial Feature Enhancement (SFE) module is employed to enhance spatial features and fuse the outputs from both branches. Building upon the SFE module, an advanced Spatial Feature Enhancement with Geometrically Adaptive Fusion (SFE-GAF) module is subsequently developed. In the SFE-GAF module, adaptive sampling is utilized to focus on true vessel morphology precisely. The adaptive process strengthens salient vascular features while significantly reducing background noise and computational overhead. Extensive experiments on the DRIVE, STARE, and CHASE_DB1 datasets validate that DB-KAUNet achieves leading segmentation performance and demonstrates exceptional robustness.

Figures (10)

• Figure 1: Common challenges in retinal fundus image vessel segmentation.
• Figure 2: An illustration of typical hybrid CNN-Transformer architectures. (a) Shallow parallel fusion of CNN and Transformer branches. (b) Parallel dual-encoder with cross-level feature interaction. (c) Serial cascade fusion: CNN output feeds Transformer block. (d) Serial bottleneck fusion: Transformer embedded in U-Net bottleneck. (e) Parallel dual-encoder with same-level feature interaction.
• Figure 3: The overall architecture of the proposed DB-KAUNet.
• Figure 4: Illustration of the HDBE and its components. (a) The structure of the HDBE. (b) The Squeeze-and-Attention Module. (c) The Position Attention Module.
• Figure 5: Detailed structure of the proposed KANConv block.