基于多维度注意力和级联解码的脊柱图像分割

doi:10.3969/j.issn.0258-8021.2025.05.002

Abstract
Figure/Table
References
Related Citation (15)

Download: PDF (3390 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract Computed Tomography (CT) technology is crucial for the diagnosis and treatment of spinal diseases. However, the complex structure and unclear boundaries of the spine in spinal CT images lead to poor segmentation accuracy. To address these issues, a spinal image segmentation network based on multi-dimensional attention and cascaded decoding was proposed in this work. First, a multi-dimensional attention module was added to the encoder, which fully extracted and fused features from multiple dimensions, thereby enhancing the model's representation ability. Second, a cascaded refinement decoding structure was constructed in the decoder. Different segmentation tasks were designed for the two decoders, which decoded the encoded features successively to refine the segmentation boundaries step by step and ensured the recovery of fine features. Finally, the proposed MACD-Net was validated on the VerSe2019 dataset (comprising 160 CT images), with Dice similarity coefficient (DSC) values reaching 91.94 %, 91.59 %, and 72.69% for the cervical, thoracic, and lumbar vertebrae, respectively, and 95^th percentile Hausdorffdistance (HD95) values of 2.24 mm, 2.93 mm, and 6.90 mm, respectively. Compared with other models, the proposed model achieved significantly better segmentation results. Additionally, experiments conducted on the CTSpine1K dataset (with 300 randomly selected CT images) further verified the effectiveness of the model. In conclusion, this study demonstrated that MACD-Net achieved precise spinal image segmentation, thereby assisting doctors in diagnosis and treatment planning.

Key words： image segmentation spine attention mechanism cascaded decoding

Received: 28 October 2024

PACS:

R318

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Mei Xiajin
	Ma Yuliang
	Zhang Wenxin
	Lv Qiang

Cite this article:

Mei Xiajin,Ma Yuliang,Zhang Wenxin, et al. MACD-Net: Spine Image Segmentation Based on Multi-Dimensional Attention and Cascaded Decoding[J]. Chinese Journal of Biomedical Engineering, 2025, 44(5): 523-532.

URL:

http://cjbme.csbme.org/EN/10.3969/j.issn.0258-8021.2025.05.002 OR http://cjbme.csbme.org/EN/Y2025/V44/I5/523

[1] 姜百浩,刘静,仇大伟,等.深度学习在脊柱图像分割中的应用综述[J].计算机工程, 2024, 50(3): 1-15.
[2] 徐云龙.多阈值与多图谱方法在医学图像分割上的应用研究[D].济南: 山东大学, 2017.
[3] 牛月凤.基于区域水平集理论的医学图像分割算法研究[D].西安: 中国科学院大学 (中国科学院西安光学精密机械研究所), 2020.
[4] 刘伟,宁静涛.一种基于像素连接性的边缘图像分割方法[J].自动化与仪器仪表, 2017 (7): 65-67.
[5] Shelhamer E, Long J, Darrell T.Fully convolutional networks for semantic segmentation[J].IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(4): 640-651.
[6] Ronneberger O, Fischer P, Brox T.U-net: convolutional networks for biomedical image segmentation[C]//18th International Conference on Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015.Munich: Springer International Publishing, 2015: 234-241.
[7] 普钟,张俊华,黄昆,等.融合多注意力机制的脊椎图像分割方法[J].计算机应用研究, 2023, 40(4): 1256-1262.
[8] 黄昆,张俊华,普钟.基于深度学习的脊椎CT图像分割[J].电子测量技术, 2022, 45(20): 151-159.
[9] 金顺楠,周迪斌,何斌,等.基于多尺度特征融合与注意力机制的脊柱CT图像分割[J].计算机系统应用, 2021, 30(10): 280-286.
[10] 田丰源,周明全,闫峰,等.基于AttentionNet和DenseUnet的脊椎CT分割[J].激光与光电子学进展, 2020, 57(20): 87-96.
[11] 毛孝鑫,宋烨,郝泳涛.基于Transformer的脊柱CT图像分割[J].电脑知识与技术, 2021, 17(20): 124-126.
[12] 杨玉聃,张俊华,刘云凤.基于三维循环残差卷积的脊柱CT图像分割[J].计算机工程, 2024, 50(4): 237-246.
[13] Meng D, Boyer E, Pujades S.Vertebrae localization, segmentation and identification using a graph optimization and an anatomic consistency cycle[J].Computerized Medical Imaging and Graphics, 2023, 107: 102235.
[14] Payer C, Stern D, Bischof H, et al.Coarse to fine vertebrae localization and segmentation with Spatial Configuration-Net and U-Net[C]//15th International Conference on Computer Vision Theory and Applications (VISAPP).Malta: SCITEPRESS, 2020: 124-133.
[15] Tao R, Liu W, Zheng G.Spine-transformers: vertebra labeling and segmentation in arbitrary field-of-view spine CTs via 3D transformers[J].Medical Image Analysis, 2022, 75: 102258.
[16] You X, Gu Y, Liu Y, et al.EG-Trans3DUNet: a single-staged transformer-based model for accurate vertebrae segmentation from spinal CT images[C]//2022 IEEE 19th International Symposium on Biomedical Imaging (ISBI).Piscataway: IEEE, 2022: 1-5.
[17] You X, Gu Y, Liu Y, et al.VerteFormer: A single-staged Transformer network for vertebrae segmentation from CT images with arbitrary field of views[J].Medical Physics, 2023, 50(10): 6296-6318.
[18] You X, He J, Yang J, et al.Learning with explicit shape priors for medical image segmentation[J].IEEE Transactions on Medical Imaging, 2024, 44(2): 927-940.
[19] Sekuboyina A, Husseini ME, Bayat A, et al.VerSe: a vertebrae labelling and segmentation benchmark for multi-detector CT images[J].Medical Image Analysis, 2021, 73: 102166.
[20] Deng Y, Wang C, Hui Y, et al.CTSpine1K: a large-scale dataset for spinal vertebrae segmentation in computed tomography[J/OL].arXiv preprint arXiv:2105.14711v4, 2021-10-03/2024-10-28.
[21] Çiçek Ö, Abdulkadir A, Lienkamp SS, et al.3D U-Net: learning dense volumetric segmentation from sparse annotation[C]//19th International Conference on Medical Image Computing and Computer-Assisted Intervention-MICCAI 2016.Athens: Springer International Publishing, 2016: 424-432.
[22] Milletari F, Navab N, Ahmadi SA.V-net: fully convolutional neural networks for volumetric medical image segmentation[C]//2016 Fourth International Conference on 3D Vision (3DV).Stanford: IEEE, 2016: 565-571.
[23] Isensee F, Jaeger PF, Kohl SAA, et al.nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation[J].Nature Methods, 2021, 18(2): 203-211.
[24] Chen J, Lu Y, Yu Q, et al.Transunet: transformers make strong encoders for medical image segmentation[J/OL].arXiv preprint arXiv:2102.04306v1, 2021-02-08/2024-10-28.
[25] Xie Y, Zhang J, Shen C, et al.Cotr: efficiently bridging cnn and transformer for 3D medical image segmentation[C]//24th International Conference on Medical Image Computing and Computer Assisted Intervention-MICCAI 2021.Strasbourg:Springer International Publishing, 2021: 171-180.
[26] Valanarasu MJ, Patel VM.Unext: Mlp-based rapid medical image segmentation network[C]//International Conference on Medical Image Computing and Computer-Assisted Intervention.Cham: Springer Nature Switzerland, 2022: 23-33.
[27] Cheng B, Schwing A, Kirillov A.Per-pixel classification is not all you need for semantic segmentation[J].Advances in Neural Information Processing Systems, 2021, 34: 17864-17875.
[28] Tang Y, Yang D, Li W, et al.Self-supervised pre-training of swin transformers for 3D medical image analysis[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.New Orleans: IEEE/CVF, 2022: 20730-20740.
[29] Islam M, Vibashan VS, Jose VJM, et al.Brain tumor segmentation and survival prediction using 3D attention UNet[C]//5th International MICCAI Brainlesion Workshop.Shenzhen: Springer International Publishing, 2020: 262-272.