| 70 | 0 | 102 |
| 下载次数 | 被引频次 | 阅读次数 |
为提高髋部双能X线吸收法(Dual-energy X-ray Absorptiometry, DXA)图像中股骨的分割精度，提出一种双分支动态蛇形卷积网络的髋部DXA图像股骨分割方法。该方法基于经典U-Net网络，由动态蛇形卷积模块、注意力机制模块和多层感知机模块构成。利用动态蛇形卷积模块的可变形卷积自适应调整感受野，深入捕捉股骨边缘细节。采用多维度注意力融合的方式设计注意力机制模块，包含位置注意力模块、通道注意力模块以及坐标注意力模块，提升模型对关键信息的关注能力，减少无关特征干扰。结合局部图像块处理与全局信息协同的联动机制，通过多层感知机模块有效融合局部与全局特征，进而定位并识别股骨相关目标区域。实验结果表明，所提方法在验证数据集上与对比方法相比，Dice系数、召回率以及交并比分别最大提高了0.86%、0.93%和1.74%,Hausdorff距离最大降低了3.96%,细节与边缘特征捕捉方面更具优势，有效提升了髋部DXA图像股骨的分割精度。
Abstract：To enhance the femoral segmentation accuracy in hip dual-energy X-ray absorptiometry(DXA) images, a femoral segmentation method for hip DXA images based on a dual-branch dynamic snake convolution network is proposed. Based on the classical U-Net, this method comprises a dynamic snake convolution module, an attention mechanism module, and a multi-layer perceptron module. The deformable convolution of the dynamic snake convolution module adaptively adjusts the receptive field to capture femoral edge details in depth. Constructed via multi-dimensional attention fusion, the attention mechanism module includes position, channel, and coordinate attention sub-modules, enhancing the model's focus on key information and reducing irrelevant feature interference. Via the linkage mechanism of local image block processing and global information coordination, the multi-layer perceptron module effectively fuses local and global features to locate and identify femoral-related target regions. Experimental results on the validation dataset demonstrate that, compared with comparative methods, the proposed method achieves maximum improvements of 0.86%, 0.93%, and 1.74% in Dice coefficient, recall rate, and intersection over union(IoU), respectively, with a maximum 3.96% reduction in Hausdorff distance. It outperforms others in capturing details and edge features, effectively enhancing the femoral segmentation accuracy for hip DXA images.
[1] Xu Y Y,Zhu Y F,Lin W L,et al.Quantitative computed tomography analysis of proximal femur bone mineral density and its relation to hip fracture risk[J].Quantitative Imaging in Medicine and Surgery,2024,14(12):9385-9393.
[2] Xu Y Y,Zhu Y F,Lin W L,et al.Quantitative computed tomography analysis of proximal femur bone mineral density and its relation to hip fracture risk[J].Quantitative Imaging in Medicine and Surgery,2024,14(12):9385-9393.
[3] Cheung W H,Hung V W,Cheuk K Y,et al.Best performance parameters of HR-pQCT to predict fragility fracture:Systematic review and meta-analysis[J].Journal of Bone and Mineral Research,2020,36(12):2381-2398.
[4] 呼伟，徐巧枝，葛湘巍，等.医学图像分割的无监督域适应研究综述[J].计算机工程与应用，2024,60(6):10-26.Hu W,Xu Q Z,Ge X W,et al.Review of unsupervised domain adaptation in medical image segmentation[J].Computer Engineering and Applications,2024,60(6):10-26.(in Chinese)
[5] 赵凤，孙磊，刘汉强.代理多目标粒子群驱动的粗糙聚类图像分割算法[J].西安邮电大学学报，2024,29(2):74-83.Zhao F,Sun L,Liu H Q.Surrogate multi-objective particle swam driven rough clustering for image segmentation[J].Journal of Xi’an University of Posts and Telecommunications,2024,29(2):74-83.(in Chinese)
[6] 兰蓉，孙宇浩，赵凤，等.基于混合反向注意力机制的息肉分割网络[J].西安邮电大学学报，2024,29(1):87-95.Lan R,Sun Y H,Zhao F,et al.Polyp segmentation network based on hybrid reverse attention mechanism[J].Journal of Xi’an University of Posts and Telecommunications,2024,29(1):87-95.(in Chinese)
[7] 孙静宜.深度学习框架下的影像辅助诊断算法设计[J].电子设计工程，2024,32(21):141-145.Sun J Y.Design of image assisted diagnosis algorithm under deep learning framework[J].Electronic Design Engineering,2024,32(21):141-145.(in Chinese)
[8] Berroukham A,Housni K,Lahraichi M.Vision transformers:A review of architecture,applications,and future directions[C]//2023 7th IEEE Congress on Information Science and Technology.Agadir - Essaouira:IEEE,2023:205-210.
[9] Wang T,Li J K,Wu H N,et al.ResLNet:Deep residual LSTM network with longer input for action recognition[J].Frontiers of Computer Science,2022,16(6):166334.
[10] Vaswani A,Shazeer N,Parmar N,et al.Attention is all you need [J].Communications of the ACM,2017,60(6):19-28.
[11] Goodfellow I,Pouget-Abadie J,Mirza M,et al.Generative adversarial networks[J].Communications of the ACM,2020,63(11):139-144.
[12] Ronneberger O,Fischer P,Brox T.U-Net:Convolutional networks for biomedical image segmentation[C]//Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015.Cham:Springer,2015:234-241.
[13] 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.(in Chinese)
[14] Guo C L,Szemenyei M,Yi Y G,et al.SA-UNet:Spatial attention U-Net for retinal vessel segmentation[C]//2020 25th International Conference on Pattern Recognition.Milan:IEEE,2021:1236-1242.
[15] Lou A G,Guan S Y,Loew M H.DC-UNet:Rethinking the U-Net architecture with dual channel efficient CNN for medical image segmentation[C]//Medical Imaging 2021:Image Processing.[S.l.]:SPIE,2021:98.
[16] Jha D,Smedsrud P H,Riegler M A,et al.ResUNet++:An advanced architecture for medical image segmentation[C]//2019 IEEE International Symposium on Multimedia.San Diego:IEEE,2019:225-2255.
[17] Ibtehaz N,Rahman M S.MultiResUNet:Rethinking the U-Net architecture for multimodal biomedical image segmentation[J].Neural Networks,2020,121:74-87.
[18] Cui J,Araujo D A,Saha S M,et al.CS-mixer:A cross-scale vision multilayer perceptron with spatial–channel mixing[J].IEEE Transactions on Artificial Intelligence,2024,5(10):4915-4927.
[19] Tolstikhin I,Houlsby N,Kolesnikov A,et al.MLP-Mixer:An all-MLP architecture for vision[EB/OL].[2025-02-03].https://doi.org/10.48550/arXiv.2105.01601.
[20] Valanarasu J M J,Patel V M.UNeXt:MLP-based rapid medical image segmentation network[C]//Medical Image Computing and Computer Assisted Intervention-MICCAI 2022.Cham:Springer,2022:23-33.
[21] Qi Y L,He Y T,Qi X M,et al.Dynamic snake convolution based on topological geometric constraints for tubular structure segmentation[C]//2023 IEEE/CVF International Conference on Computer Vision.Paris:IEEE,2023:6047-6056.
[22] Wang C L,Ning H J,Chen X R,et al.DB-UNet:MLP based dual branch UNet for accurate vessel segmentation in OCTA images[C]//ICASSP 2023-2023 IEEE International Conference on Acoustics,Speech and Signal Processing.Rhodes Island:IEEE,2023:1-5.
[23] Fu J,Liu J,Tian H J,et al.Dual attention network for scene segmentation[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Long Beach:IEEE,2019:3141-3149.
[24] Hou Q B,Zhou D Q,Feng J S.Coordinate attention for efficient mobile network design[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Nashville:IEEE,2021:13708-13717.
[25] Wu R K,Liu Y H,Ning G C,et al.UltraLight VM-UNet:Parallel Vision Mamba significantly reduces parameters for skin lesion segmentation[J].Patterns,2025,6(11):101298.
[26] Mumuni A,Mumuni F.Data augmentation:A comprehensive survey of modern approaches[J].Array,2022,16:100258.
[27] 钟乔鑫，赵毅忠，张飞燕，等.基于模态交互学习的多源心脏图像分割方法研究[J].磁共振成像，2024,15(4):145-152.Zhong Q X,Zhao Y Z,Zhang F Y,et al.Research on multi-source cardiac image segmentation method based on modal interaction learning[J].Chinese Journal of Magnetic Resonance Imaging,2024,15(4):145-152.(in Chinese)
[28] Yacouby R,Axman D.Probabilistic extension of precision,recall,and F1 score for more thorough evaluation of classification models[C]//First Workshop on Evaluation and Comparison of NLP Systems.Stroudsburg:ACL,2020:79-91.
[29] Shelhamer E,Long J,Darrell T.Fully convolutional networks for semantic segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(4):640-651.
[30] Haddad S,Halder A.A note on the Hausdorff distance between norm balls and their linear maps[J].Set-Valued and Variational Analysis,2023,31(3):30.
基本信息:
DOI：10.13682/j.issn.2095-6533.2026.02.010
中图分类号:R816.8;TP391.41
引用信息:
[1]魏秋月,姚可,谈伟,等.双分支动态蛇形卷积网络的髋部DXA图像股骨分割方法[J].西安邮电大学学报,2026,31(02):100-108.DOI:10.13682/j.issn.2095-6533.2026.02.010.
基金信息:
陕西省自然科学基础研究计划项目(2023-JC-YB-521); 自贡市哲学社会科学重点研究基地运动与健康创新研究中心重点课题(YD-JKZ23-05)
2026-03-04
2026-03-04
2026-03-04