SA-UMamba: Spatial attention convolutional neural networks for medical image segmentation.
Medical image segmentation plays an important role in medical diagnosis and treatment. Most recent medical image segmentation methods are based on a convolutional neural network (CNN) or Transformer model. However, CNN-based methods are limited by locality, whereas Transformer-based methods are cons...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Public Library of Science (PLoS)
2025-01-01
|
| Series: | PLoS ONE |
| Online Access: | https://doi.org/10.1371/journal.pone.0325899 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850100415064965120 |
|---|---|
| author | Lei Liu Zhao Huang Shuai Wang Jun Wang Baosen Liu |
| author_facet | Lei Liu Zhao Huang Shuai Wang Jun Wang Baosen Liu |
| author_sort | Lei Liu |
| collection | DOAJ |
| description | Medical image segmentation plays an important role in medical diagnosis and treatment. Most recent medical image segmentation methods are based on a convolutional neural network (CNN) or Transformer model. However, CNN-based methods are limited by locality, whereas Transformer-based methods are constrained by the quadratic complexity of attention computations. Alternatively, the state-space model-based Mamba architecture has garnered widespread attention owing to its linear computational complexity for global modeling. However, Mamba and its variants are still limited in their ability to extract local receptive field features. To address this limitation, we propose a novel residual spatial state-space (RSSS) block that enhances spatial feature extraction by integrating global and local representations. The RSSS block combines the Mamba module for capturing global dependencies with a receptive field attention convolution (RFAC) module to extract location-sensitive local patterns. Furthermore, we introduce a residual adjust strategy to dynamically fuse global and local information, improving spatial expressiveness. Based on the RSSS block, we design a U-shaped SA-UMamba segmentation framework that effectively captures multi-scale spatial context across different stages. Experiments conducted on the Synapse, ISIC17, ISIC18 and CVC-ClinicDB datasets validate the segmentation performance of our proposed SA-UMamba framework. |
| format | Article |
| id | doaj-art-0ce381ccebb443febefa6690319febda |
| institution | DOAJ |
| issn | 1932-6203 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Public Library of Science (PLoS) |
| record_format | Article |
| series | PLoS ONE |
| spelling | doaj-art-0ce381ccebb443febefa6690319febda2025-08-20T02:40:18ZengPublic Library of Science (PLoS)PLoS ONE1932-62032025-01-01206e032589910.1371/journal.pone.0325899SA-UMamba: Spatial attention convolutional neural networks for medical image segmentation.Lei LiuZhao HuangShuai WangJun WangBaosen LiuMedical image segmentation plays an important role in medical diagnosis and treatment. Most recent medical image segmentation methods are based on a convolutional neural network (CNN) or Transformer model. However, CNN-based methods are limited by locality, whereas Transformer-based methods are constrained by the quadratic complexity of attention computations. Alternatively, the state-space model-based Mamba architecture has garnered widespread attention owing to its linear computational complexity for global modeling. However, Mamba and its variants are still limited in their ability to extract local receptive field features. To address this limitation, we propose a novel residual spatial state-space (RSSS) block that enhances spatial feature extraction by integrating global and local representations. The RSSS block combines the Mamba module for capturing global dependencies with a receptive field attention convolution (RFAC) module to extract location-sensitive local patterns. Furthermore, we introduce a residual adjust strategy to dynamically fuse global and local information, improving spatial expressiveness. Based on the RSSS block, we design a U-shaped SA-UMamba segmentation framework that effectively captures multi-scale spatial context across different stages. Experiments conducted on the Synapse, ISIC17, ISIC18 and CVC-ClinicDB datasets validate the segmentation performance of our proposed SA-UMamba framework.https://doi.org/10.1371/journal.pone.0325899 |
| spellingShingle | Lei Liu Zhao Huang Shuai Wang Jun Wang Baosen Liu SA-UMamba: Spatial attention convolutional neural networks for medical image segmentation. PLoS ONE |
| title | SA-UMamba: Spatial attention convolutional neural networks for medical image segmentation. |
| title_full | SA-UMamba: Spatial attention convolutional neural networks for medical image segmentation. |
| title_fullStr | SA-UMamba: Spatial attention convolutional neural networks for medical image segmentation. |
| title_full_unstemmed | SA-UMamba: Spatial attention convolutional neural networks for medical image segmentation. |
| title_short | SA-UMamba: Spatial attention convolutional neural networks for medical image segmentation. |
| title_sort | sa umamba spatial attention convolutional neural networks for medical image segmentation |
| url | https://doi.org/10.1371/journal.pone.0325899 |
| work_keys_str_mv | AT leiliu saumambaspatialattentionconvolutionalneuralnetworksformedicalimagesegmentation AT zhaohuang saumambaspatialattentionconvolutionalneuralnetworksformedicalimagesegmentation AT shuaiwang saumambaspatialattentionconvolutionalneuralnetworksformedicalimagesegmentation AT junwang saumambaspatialattentionconvolutionalneuralnetworksformedicalimagesegmentation AT baosenliu saumambaspatialattentionconvolutionalneuralnetworksformedicalimagesegmentation |