| 186 | 1 | 44 |
| Downloads | Citas | Reads |
In order to solve the problem that spine magnetic resonance image edge blurring and pixel distribution imbalance affecting the accurate segmentation of the spine image, an edge attention U-Net method was proposed to enhance the extraction of edge features. The edge attention module was constructed using the multi-head self-attention mechanism and the convolutional attention mechanism. And the local features and global representations of the edges were fused to enhance the learning of edge features. Then, the focus category loss function was proposed to reduce the effect of imbalanced pixel distribution. The corresponding weight factors were adjusted according to the loss values of different pixel regions, so that the network could pay attention to pixel regions containing fine feature structures. The experimental results showed that compared with other segmentation networks, edge attention U-Net improved the segmentation accuracy, and the segmentation performance of the network trained with focus category loss function was better than other loss functions with the same network conditions, which proved the effectiveness of the proposed approach.
[1] LIU X B,SONG L P,LIU S,et al.A review of deep-learning-based medical image segmentation methods[J].Sustainability,2021,13(3):1224.
[2] RONNEBERGER O,FISCHER P,BROX T.U-Net:convolutional networks for biomedical image segmentation[C]//Proceedings of the Medical Image Computing and Computer-assisted Intervention.Cham:Springer International Publishing,2015:234-241.
[3] 崔珂，田启川，廉露.基于U-Net变体的医学图像分割算法综述[J].计算机工程与应用，2024,60(11):32-49.CUI K,TIAN Q C,LIAN L.Review of medical image segmentation algorithms based on U-Net variants[J].Computer engineering and applications,2024,60(11):32-49.
[4] OKTAY O,SCHLEMPER J,FOLGOC L,et al.Attention U-Net:learning where to look for the pancreas[EB/OL].(2018-05-20)[2024-11-20].https://doi.org/10.48550/arXiv.1804.03999.
[5] ZHOU Z W,SIDDIQUEE M M R,TAJBAKHSH N,et al.UNet++:a nested U-Net architecture for medical image segmentation[C]// Proceedings of the Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support.Cham:Springer International Publishing,2018:3-11.
[6] VASWANI A,SHAZEER N,PARMAR N,et al.Attention is all you need [EB/OL].(2023-08-02) [2024-10-25].https://doi.org/10.48550/arXiv.1706.03762.
[7] 肖斌，罗浩，张恒宾，等.多尺度融合卷积的轻量化Transformer无人机地物识别模型[J].郑州大学学报(理学版),2024,56(1):32-39.XIAO B,LUO H,ZHANG H B,et al.A lightweight Transformer UAV surface feature recognition model based on multi-scale fusion convolutional networks[J].Journal of Zhengzhou university (natural science edition),2024,56(1):32-39.
[8] CHEN J N,MEI J R,LI X H,et al.TransUNet:rethinking the U-Net architecture design for medical image segmentation through the lens of transformers[J].Medical image analysis,2024,97:103280.
[9] CAO H,WANG Y Y,CHEN J,et al.Swin-Unet:Unet-like pure transformer for medical image segmentation[C]//European Conference on Computer Vision.Cham:Springer International Publishing,2023:205-218.
[10] LIU Z,LIN Y T,CAO Y,et al.Swin Transformer:hierarchical vision transformer using shifted windows[C]//IEEE/CVF International Conference on Computer Vision.Piscataway:IEEE Press,2021:9992-10002.
[11] ZHANG Z J,FU H Z,DAI H,et al.ET-Net:a generic edge-attention guidance network for medical image segmentation[C]//International Conference on Medical Image Computing and Computer Assisted Intervention.Cham:Springer International Publishing,2019:442-450.
[12] GUO J N,ZENG W,YU S,et al.RAU-Net:U-Net model based on residual and attention for kidney and kidney tumor segmentation[C]//IEEE International Conference on Consumer Electronics and Computer Engineering.Piscataway:IEEE Press,2021:353-356.
[13] 陈俊松，易积政，陈爱斌.CNS-Net:一种循环多注意力特征聚合架构[J].郑州大学学报(理学版),2024,56(2):73-79.CHEN J S,YI J Z,CHEN A B.CNS-Net:a cyclic multi-attention feature aggregation architecture[J].Journal of Zhengzhou university (natural science edition),2024,56(2):73-79.
[14] HE S Y,LI Q,LI X D,et al.A lightweight convolutional neural network based on dynamic level-set loss function for spine MR image segmentation[J].Journal of magnetic resonance imaging,2024,59(4):1438-1453.
[15] LIN T Y,GOYAL P,GIRSHICK R,et al.Focal loss for dense object detection[C]//IEEE International Conference on Computer Vision.Piscataway:IEEE Press,2017:2999-3007.
[16] RUAN J C,LI J C,XIANG S C.VM-UNet:vision mamba UNet for medical image segmentation[EB/OL].(2024-11-08)[2024-11-20].https://doi.org/10.48550/arXiv.2402.02491.
[17] RUAN J C,XIE M Y,GAO J S,et al.EGE-UNet:an efficient group enhanced UNet for skin lesion segmentation[C]//Proceedings of the Medical Image Computing and Computer Assisted Intervention.Cham:Springer International Publishing,2023:481-490.
[18] ZHAO R J,QIAN B Y,ZHANG X L,et al.Rethinking Dice loss for medical image segmentation[C]//IEEE International Conference on Data Mining.Piscataway:IEEE Press,2020:851-860.
[19] BERTELS J,EELBODE T,BERMAN M,et al.Optimizing the Dice score and Jaccard index for medical image segmentation:theory and practice[C]//International Conference on Medical Image Computing and Computer Assisted Intervention.Cham:Springer International Publishing,2019:92-100.
Basic Information:
DOI：10.13705/j.issn.1671-6841.2024204
China Classification Code:R681.5;TP391.41
Citation Information:
[1]LI Qi,DONG Jiayan,WU Yan ,et al.Spine Image Segmentation Method Combining Edge Attention U-Net and Focus Category Loss[J].Journal of Zhengzhou University(Natural Science Edition),2026,58(04):36-43.DOI:10.13705/j.issn.1671-6841.2024204.
Fund Information:
吉林省科技发展计划项目(20230203098SF); 中山市社会公益科技研究项目(2023B2015)
2025-07-14
2025-07-14
2025-07-14