Scholarlink

<p>The combination of convolutional neural networks (CNNs) and vision transformer (ViT) in U-shaped networks offers great potential for medical image segmentation. To address the lack of explicit spatial priors in CNN–transformer architectures, we propose a novel network named RaEUNet, integrating retentive networks meet vision transformers (RMT) and modified efficient multi-scale convolutional attention decoding. RMT with explicit spatial priors replaces ViT to enhance spatial features in the encoder. For decoder, we propose the efficient channel attention up-convolution block, which reduces the impact of insensitivity to channel relationships caused by depth-wise convolution by introducing an efficient channel attention module. Additionally, we design multi-scale re-weighted attention module by introducing spatial and channel reconstruction convolution and our proposed re-weighted adjustment module, which fully integrates high-level and low-level features, reducing feature redundancy and improving information capture. Experiments on Synapse, Polyp, ACDC, and ISIC2018 datasets show superior performance over state-of-the-art methods.</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

RaEUNet: a retentive and efficient UNet for medical image segmentation

  • 期刊论文

  • 发表日期：

    2025年5月27日

查看全文
  • Yiyi Wang,
  • Jia Su,
  • Song Zhang,
  • Eisei Nakahara

摘要

The combination of convolutional neural networks (CNNs) and vision transformer (ViT) in U-shaped networks offers great potential for medical image segmentation. To address the lack of explicit spatial priors in CNN–transformer architectures, we propose a novel network named RaEUNet, integrating retentive networks meet vision transformers (RMT) and modified efficient multi-scale convolutional attention decoding. RMT with explicit spatial priors replaces ViT to enhance spatial features in the encoder. For decoder, we propose the efficient channel attention up-convolution block, which reduces the impact of insensitivity to channel relationships caused by depth-wise convolution by introducing an efficient channel attention module. Additionally, we design multi-scale re-weighted attention module by introducing spatial and channel reconstruction convolution and our proposed re-weighted adjustment module, which fully integrates high-level and low-level features, reducing feature redundancy and improving information capture. Experiments on Synapse, Polyp, ACDC, and ISIC2018 datasets show superior performance over state-of-the-art methods.