| 18 | 1 | 0 |
 Reads |
 Downloads |
 Citas |
The existing 3D U⁃Net model has high computational complexity and poor accuracy when processing 3D medical image data, so it is difficult to meet the needs of clinical real⁃time diagnosis. Therefore, a lightweight technology architecture based on 3D U⁃Net combined with Mamba is proposed to reduce the computational complexity and improve the segmentation accuracy of the traditional deep learning models in multi⁃modal brain tumor MRI image segmentation. Mamba′s linear complexity self ⁃ attention mechanism is used to optimize the 3D U⁃Net model to reduce the computational complexity and improve the inference speed and segmentation accuracy. The specific method is to introduce LWM module and attention bridge module on the basis of 3D U⁃Net model. Parallel Mamba architecture is adopted in the LWM module, which reduces the number of model parameters by reducing the number of input channels, greatly reduces the computational complexity and memory consumption while extracting global information, and realizes efficient segmentation in the case of fewer parameters. In addition, the LWM further improves the segmentation performance of fine⁃grained tumor cores by introducing SE module. The experiments on the dataset BraTS2020 show that the proposed method significantly improves the accuracy of tumor segmentation, and greatly reduces the training and inference duration, so it shows great practical application potential.
[1] Çiçek Ö, Abdulkadir A, Lienkamp S S, et al. 3D U ⁃ Net: Learning dense volumetric segmentation from sparse annotation [C]// Proceedings of 19th International Conference on Medical Image Computing and Computer⁃Assisted Intervention. Athens, Greece: Springer, 2016: 424⁃432.
[2] Oktay O, Schlemper J, Folgoc L L, et al. Attention U ⁃ Net: Learning where to look for the pancreas [EB/OL]. [2018⁃05⁃20]. https://arxiv.org/abs/1804.03999.
[3] Chen Jieneng, Lu Yongyi, Yu Qihang, et al. TransUNet: Transformers make strong encoders for medical image segmentation [EB/OL]. [2021 ⁃ 04 ⁃ 12]. https://arxiv. org/abs/ 2102.04306.
[4] Cao Hu, Wang Yueyue, Chen J, et al. Swin⁃Unet: Unet⁃like pure transformer for medical image segmentation [C]// International Conference on Medical Image Computing and Computer ⁃ Assisted Intervention (MICCAI). Tel Aviv, Israel: Springer, 2022: 205⁃218.
[5] Wang Wenxuan, Chen Chen, Ding Meng, et al. TransBTS: Multimodal brain tumor segmentation using transformer [C]// Proceedings of 24th International Conference on Medical Image Computing and Computer Assisted Intervention. Strasbourg, France: Springer, 2021: 109⁃119.
[6] Isensee F, Jaeger P F, Kohl S A A, et al. nnU⁃Net: A self⁃configuring method for deep learning⁃based biomedical image segmentation [J]. Nature Methods, 2021, 18(2): 203⁃211.
[7] Wang Ziyang, Zheng Jianqing, Zhang Yichi, et al. Mamba⁃UNet: UNet⁃like pure visual mamba for medical image segmentation [EB/OL]. [2025 ⁃ 05 ⁃ 31]. https://doi. org/10.48550/ arXiv.2402.05079.
[8] Roy S K, Dubey S R, Chatterjee S, et al. FuSENet: Fused squeeze ⁃ and ⁃ excitation network for spectral ⁃ spatial hyperspectral image classification [J]. IET Image Processing, 2020, 14(8): 1653⁃1661.
[9] Zhong Xian, Gong Oubo, Huang Wenxin, et al. Squeeze⁃and⁃ excitation wide residual networks in image classification [C]// 2019 IEEE International Conference on Image Processing (ICIP). [S.l.]: IEEE, 2019: 395⁃399.
[10] Deng Jian, Ma Yanyun, Li Deng’ao, et al. Classification of breast density categories based on SE ⁃ Attention neural networks [J]. Computer Methods and Programs in Biomedicine, 2020, 193: 105489.
[11] Gu A, Dao T. Mamba: Linear⁃time sequence modeling with selective state spaces [EB/OL]. [2025⁃01⁃19]. https://doi.org/ 10.48550/arXiv.2312.00752.
[12] Ruan Jiacheng, Xiang Suncheng, Xie Mingye, et al. MALUNet: A multi ⁃ attention and light ⁃ weight UNet for skin lesion segmentation [C]// IEEE International Conference on Bioinformatics and Biomedicine. Las Vegas, NV, USA: IEEE, 2022: 1150⁃1156.
[13] Woo S, Park J, Lee J Y, et al. CBAM: Convolutional block attention module [C]// Proceedings of European Conference on Computer Vision. Munich, Germany: Springer, 2018: 3⁃19.
Basic Information:
DOI：10.16652/j.issn.1004⁃373x.2026.09.006
Citation Information:
[1]Hou Xiangning12, Huang Xiaobin12, Xu Caocao12 ,et al.Lightweight multi⁃modal brain tumor MRI image segmentation based on Mamba[J].Modern Electronic Technique,2026(9):32-37.DOI:10.16652/j.issn.1004⁃373x.2026.09.006.
Fund Information:
乐山市科技局基金项目（24YYJC0004）; 成都理工大学工程技术学院基金项目（C122024002）