基于各向异性注意力的双分支血管分割模型

doi:10.19678/j.issn.1000-3428.0067078

摘要/Abstract

摘要：

血管分割对于血管疾病的诊断和治疗具有重要意义，但由于血管边界模糊、病变血管的形状多变且不同样本之间的差异性较大，因此要求分割模型能够准确地挖掘血管与背景类间的差异性以及血管内部的连通性。提出一种基于中心线约束与各向异性注意力的新型三维血管分割网络CAU-Net。针对血管分割的难点，对基础网络结构ResU-Net进行改进，构建各向异性注意力模块，该模块根据管腔结构特有的空间各向异性，从3个方向提取血管空间各向异性特征，并对特征通道间的相关性进行建模，学习血管的三维空间信息。采用主-辅双分支模型，b-Net对血管进行语义分割，a-Net学习血管中心线的连续性特征，约束b-Net的血管分割结果，保证血管分割结果的完整性。在公开数据集3D-IRCADb-01上的实验结果表明，对于门静脉及肝静脉的分割，CAU-Net分别取得(74.80±8.05）%和(76.14±6.89)%的Dice系数、(54.80±8.09)%和(50.40±5.22)%的NSD系数、(72.43±8.26)%和(70.84±6.05)%的clDice系数、(46.47±12.89)%和(39.19±7.97)%的分支检测率以及(67.08±15.59)%和(61.47±9.32)%的树长检测率。在公开脑血管数据集IXI上进行组件消融实验，模型在验证集上的平均Dice、NSD、clDice、BD和TD分别为（94.11±0.39）%、（96.53±0.37）%、（95.83±0.59）%、（98.64±1.63）%和（95.44±1.22）%，相比于Baseline分别提升了0.92%、0.82%、0.92%、1.11%和1.60%。CAU-Net血管分割模型能够显著提升血管分割的精度和完整度。

关键词: 血管分割, 中心线约束, 各向异性, 注意力机制, 双分支模型

Abstract:

Vascular segmentation is significant for diagnosing and treating vascular diseases. However, because of the fuzzy boundary of vessels, the variable shape of diseased vessels, and the significant differences between different samples, the segmentation model should accurately determine the differences between vessels and background classes and analyze the connectivity within vessels. This study proposes a novel three-dimensional vascular segmentation network, CAU-Net, based on centerline constraints and anisotropic attention. In response to the difficulties in vascular segmentation, the basic network structure, ResU-Net, is improved to construct an anisotropic attention module. This module extracts vascular spatial anisotropic features from three directions based on the unique spatial anisotropy of the vascular structure and models the correlation between feature channels to learn the three-dimensional spatial information of the vessels. By using the main auxiliary dual-branch model, b-Net performs semantic segmentation on vessels, whereas a-Net learns the continuity features of vessel centerlines, constrains the vascular segmentation results of b-Net, and ensures the integrity of the vascular segmentation results. The experimental results on the publicly available dataset 3D-IRCADb-01 shows that for the segmentation of portal and hepatic veins, CAU-Net achieves Dice coefficients of (74.80±8.05)% and (76.14±6.89)%, NSD coefficients of(54.80±8.09)% and(50.40±5.22)%, clDice coefficients of (72.43±8.26)% and(70.84±6.05)%, Branch Detection(BD) rates of (46.47±12.89)% and(39.19±7.97)%, and Tree length Detection(TD) rates of(67.08±15.59)% and(61.47±9.32)%, respectively. Component ablation experiments are conducted on the publicly available cerebrovascular dataset IXI, and the average Dice, NSD, clDice, BD, and TD values of the model on the validation set are(94.11±0.39)%, (96.53±0.37)%, (95.83±0.59)%, (98.64±1.63)%, and(95.44±1.22)%, respectively. Compared to the Baseline, the average Dice, NSD, clDice, BD, and TD values of the proposed model increased by 0.92%, 0.82%, 0.92%, 1.11%, and 1.60%, respectively. The CAU-Net vascular segmentation model can significantly improve the accuracy and completeness of vascular segmentation.

Key words: vascular segmentation, centerline constraint, anisotropy, attention mechanism, dual-branch model

徐晓峰, 黄韫栀, 徐军. 基于各向异性注意力的双分支血管分割模型[J]. 计算机工程, 2024, 50(1): 348-356.

Xiaofeng XU, Yunzhi HUANG, Jun XU. Dual-Branch Vascular Segmentation Model Based on Anisotropic Attention[J]. Computer Engineering, 2024, 50(1): 348-356.

http://www.ecice06.com/CN/Y2024/V50/I1/348

图/表 10

图1 血管分割中的挑战性问题示例

Fig.1 Examples of challenging issues in vascular segmentation

图2 CAU-Net总体结构

Fig.2 Overall structure of CAU-Net

图3 b-Net模型结构

Fig.3 Structure of the b-Net model

图4 a-Net模型结构

Fig.4 Structure of the a-Net model

图5 公开数据集血管三维可视化图

Fig.5 Three-dimensional visualization of vessels from a publicly available dataset

图6 肝脏门静脉和肝静脉分割结果

Fig.6 Segmentation results of the hepatic portal vein and hepatic vein

图7 脑血管分割结果对比

Fig.7 Comparison of the segmentation results of the cerebrovascular

参考文献 30

1	KREATSOULAS C, ANAND S S. The impact of social determinants on cardiovascular disease. Canadian Journal of Cardiology, 2010, 26 (Suppl. C): 8- 13.
2	SMITH-BINDMAN R, KWAN M L, MARLOW E C, et al. Trends in use of medical imaging in US health care systems and in Ontario, Canada, 2000-2016. JAMA, 2019, 322 (9): 843- 856. doi: 10.1001/jama.2019.11456
3	FORUZAN AMIR H, ZOROOFI REZA A, YOSHINOBU S, et al. A Hessian-based filter for vascular segmentation of noisy hepatic CT scans. International Journal of Computer Assisted Radiology and Surgery, 2012, 7 (2): 199- 205. doi: 10.1007/s11548-011-0640-y
4	TIAN Y, CHEN Q L, WANG W, et al. A vessel active contour model for vascular segmentation. BioMed Research International, 2014, 2014, 1- 15. doi: 10.1007/s00138-016-0805-3
5	FRIMAN O, HINDENNACH M, KÜHNEL C, et al. Multiple hypothesis template tracking of small 3D vessel structures. Medical Image Analysis, 2010, 14 (2): 160- 171. doi: 10.1016/j.media.2009.12.003
6	CHI Y L, LIU J M, VENKATESH S K, et al. Segmentation of liver vasculature from contrast enhanced CT images using context-based voting. IEEE Transactions on Biomedical Engineering, 2011, 58 (8): 2144- 2153. doi: 10.1109/TBME.2010.2093523
7	RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]//Proceedings of International Conference on Medical Image Computing and Computer-Assisted Intervention. Berlin, Germany: Springer, 2015: 234-241.
8	MILLETARI F, NAVAB N, AHMADI S A. V-Net: fully convolutional neural networks for volumetric medical image segmentation[C]//Proceedings of the 14th International Conference on 3D Vision. Washington D. C., USA: IEEE Press, 2016: 565-571.
9	ZHOU Z W, RAHMAN SIDDIQUEE M M, TAJBAKHSH N, et al. UNet++: a nested U-Net architecture for medical image segmentation[C]//Proceedings of International Workshop on Deep Learning in Medical Image Analysis. Berlin, Germany: Springer, 2018: 3-11.
10	ISENSEE F, JAEGER P F, KOHL S A A, et al. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nature Methods, 2021, 18 (2): 203- 211. doi: 10.1038/s41592-020-01008-z
11	梅旭璋, 江红, 孙军. 基于密集注意力网络的视网膜血管图像分割. 计算机工程, 2020, 46 (3): 267-272, 279. URL
	MEI X Z, JIANG H, SUN J. Retinal vessel image segmentation based on dense attention network. Computer Engineering, 2020, 46 (3): 267-272, 279. URL
12	邓小波, 刘奇, 陈曦, 等. 基于通道注意力双路径架构网络分割视网膜血管. 中国医学影像技术, 2021, 37 (10): 1543- 1547. doi: 10.13929/j.issn.1003-3289.2021.10.026
	DENG X B, LIU Q, CHEN X, et al. Retinal vessel segmentation based on dual-path channel attention network. Chinese Journal of Medical Imaging Technology, 2021, 37 (10): 1543- 1547. doi: 10.13929/j.issn.1003-3289.2021.10.026
13	刘娜, 汪国强. SAF-Net: 自注意力融合网络的视网膜血管分割. 计算机工程与应用, 2023, 59 (14): 217- 223. doi: 10.3778/j.issn.1002-8331.2205-0166
	LIU N, WANG G Q. SAF-Net: self-attention fusion network for retinal vessel segmentation. Computer Engineering and Applications, 2023, 59 (14): 217- 223. doi: 10.3778/j.issn.1002-8331.2205-0166
14	TETTEH G, EFREMOV V, FORKERT N D, et al. DeepVesselNet: vessel segmentation, centerline prediction, and bifurcation detection in 3-D angiographic volumes. Frontiers in Neuroscience, 2020, 14, 592352. doi: 10.3389/fnins.2020.592352
15	KITRUNGROTSAKUL T, HAN X H, IWAMOTO Y, et al. VesselNet: a deep convolutional neural network with multi pathways for robust hepatic vessel segmentation. Computerized Medical Imaging and Graphics, 2019, 75, 74- 83. doi: 10.1016/j.compmedimag.2019.05.002
16	HUANG Q, SUN J F, DING H, et al. Robust liver vessel extraction using 3D U-Net with variant dice loss function. Computers in Biology and Medicine, 2018, 101, 153- 162. doi: 10.1016/j.compbiomed.2018.08.018
17	YAN Q S, WANG B, ZHANG W, et al. Attention-guided deep neural network with multi-scale feature fusion for liver vessel segmentation. IEEE Journal of Biomedical and Health Informatics, 2021, 25 (7): 2629- 2642. doi: 10.1109/JBHI.2020.3042069
18	KESHWANI D, KITAMURA Y, IHARA S, et al. TopNet: topology preserving metric learning for vessel tree reconstruction and labelling[EB/OL]. [2023-02-01]. https://arxiv.org/abs/2009.08674.
19	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2016: 770-778.
20	WU Y, HE K M. Group normalization[C]//Proceedings of European Conference on Computer Vision. Berlin, Germany: Springer, 2018: 3-19.
21	HU J E, SHEN L, SUN G. Squeeze-and-excitation networks[C]//Proceedings of Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2018: 7132-7141.
22	ZHU W T, HUANG Y F, ZENG L, et al. AnatomyNet: deep learning for fast and fully automated whole-volume segmentation of head and neck anatomy. Medical Physics, 2019, 46 (2): 576- 589. doi: 10.1002/mp.13300
23	KERVADEC H, BOUCHTIBA J, DESROSIERS C, et al. Boundary loss for highly unbalanced segmentation. Medical Image Analysis, 2021, 67, 101851. doi: 10.1016/j.media.2020.101851
24	SOLER L, HOSTETTLER A, AGNUS V, et al. 3D image reconstruction for comparison of algorithm database: a patient specific anatomical and medical image database[EB/OL]. [2023-02-01]. https://www.ircad.fr/research/data-sets/liver-segmentation-3d-ircadb-01,2010.
25	BIZJAK Ž, CHIEN A C, BURNIK I, et al. Novel dataset and evaluation of state-of-the-art vessel segmentation methods[C]//Proceedings of SPIE Medical Imaging. New York, USA: [s. n.], 2022: 772-780.
26	SHIT S, PAETZOLD J C, SEKUBOYINA A, et al. clDice-a novel topology-preserving loss function for tubular structure segmentation[C]//Proceedings of the Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2021: 16560-16569.
27	NIKOLOV S, BLACKWELL S, ZVEROVITCH A, et al. Deep learning to achieve clinically applicable segmentation of head and neck anatomy for radiotherapy[EB/OL]. [2023-02-01]. https://arxiv.org/pdf/1809.04430.pdf.
28	LO P, VAN GINNEKEN B, REINHARDT J M, et al. Extraction of airways from CT (EXACT'09). IEEE Transactions on Medical Imaging, 2012, 31 (11): 2093- 2107. doi: 10.1109/TMI.2012.2209674
29	XU W D, LIU H, WANG X D, et al. Liver segmentation in CT based on ResUNet with 3D probabilistic and geometric post process[C]//Proceedings of the 4th International Conference on Signal and Image Processing. Washington D. C., USA: IEEE Press, 2019: 685-689.
30	OKTAY O, SCHLEMPER J, FOLGOC L L, et al. Attention U-Net: learning where to look for the pancreas[EB/OL]. [2023-02-01]. https://arxiv.org/pdf/1804.03999.pdf.

[1]	圣文顺, 余熊峰, 林佳燕, 陈欣. 融合注意力与特征金字塔的小尺度目标检测算法[J]. 计算机工程, 2024, 50(1): 242-250.
[2]	吴志强, 解庆, 李琳, 刘永坚. 基于多模态融合的图神经网络推荐算法[J]. 计算机工程, 2024, 50(1): 91-100.
[3]	申秀雨, 姬伟峰, 李映岐, 吴玄. 面向边缘计算的TCA1C DDoS检测模型[J]. 计算机工程, 2024, 50(1): 198-205.
[4]	杨瑞君, 秦晋京, 程燕. 基于生成对抗网络的自然场景低照度增强模型[J]. 计算机工程, 2024, 50(1): 279-288.
[5]	杨静, 陆铭华, 马洁琼, 吴金平, 刘星璇. 基于交替循环神经网络的水下防御态势预测方法[J]. 计算机工程, 2023, 49(9): 69-78.
[6]	孙龙, 张荣芬, 刘宇红, 饶庭漓. 监控视角下密集人群口罩佩戴检测算法[J]. 计算机工程, 2023, 49(9): 313-320.
[7]	苏晓东, 李世洲, 赵佳圆, 亮洪宇, 张玉荣, 徐红岩. 基于多级叠加和注意力机制的图像语义分割[J]. 计算机工程, 2023, 49(9): 265-271, 278.
[8]	韩璐, 霍纬纲, 张永会, 刘涛. 基于多尺度特征融合与双注意力机制的多元时间序列预测[J]. 计算机工程, 2023, 49(9): 99-108.
[9]	龙玉江, 卫薇, 舒彧, 张正刚, 王道累, 李峰. 基于自适应关键点的破损旋转绝缘子检测方法[J]. 计算机工程, 2023, 49(9): 272-278.
[10]	丰芳宇, 罗晓曙, 蒙志明, 王广宇. 基于抗混叠残差注意力网络的人脸表情识别[J]. 计算机工程, 2023, 49(8): 190-198.
[11]	王书朋, 何引弟. 融合特征注意力机制的非均匀光照图像增强算法[J]. 计算机工程, 2023, 49(8): 232-239.
[12]	刘昊鑫, 董超, 勾智楠, 高凯. 融合混合表征的小样本关系抽取方法[J]. 计算机工程, 2023, 49(8): 63-68.
[13]	杨长沛, 廖列法. 基于门控空洞卷积特征融合的中文命名实体识别[J]. 计算机工程, 2023, 49(8): 85-95.
[14]	刘俊豪, 王美林, 谢兴, 宋烨兴, 许莉花. 基于改进YOLOv5的皮革瑕疵检测算法[J]. 计算机工程, 2023, 49(8): 240-249.
[15]	马娜, 温廷新, 贾旭, 李晓会. 复杂光照条件下自适应的车脸重识别模型[J]. 计算机工程, 2023, 49(8): 275-282, 290.

选择文件类型/文献管理软件名称

选择包含的内容