用于颈椎MRI分割的多尺度特征融合注意力网络模型

doi:10.19678/j.issn.1000-3428.0065772

摘要/Abstract

摘要：

近年来，基于深度学习的医学图像辅助诊断逐渐成为主流，但常见的医疗锥体分割模型缺乏对颈椎细节信息的提取，导致锥体分割不完整或边缘相对模糊。为了提高颈椎MRI图像的分割精度，基于ResNet构建一种多尺度特征融合注意力（MSFFA）网络模型。利用多尺度注意力模块融合不同感受野进行注意力特征增强，同时为了降低特征信息融合的损耗，采用跨尺度特征融合模块进行相似域和边缘域特征增强，最终将原始样本的特征信息整合到分割结果中进行细节增强，进一步优化模型分割性能。实验结果表明，MSFFA模型相比于U-Net、AttUNet等模型分割得到的颈椎结构更完整、边缘更平滑，同时在腰椎分割中也能取得更精确的分割结果，并且相比于最优对照模型DeepLabv3+，Dice相似系数的均值提升了1.05个百分点。

关键词: 颈椎分割, 注意力机制, 多尺度融合, 特征增强, 卷积神经网络

Abstract:

In recent years, deep learning has been increasingly employed to effectively assist the diagnosis of medical images. However, common medical cone segmentation models cannot extract cervical spine details, resulting in incomplete cone segmentation or relatively blurred edges. To improve the segmentation accuracy of cervical vertebrae Magnetic Resonance Imaging(MRI), a Multi-Scale Feature Fusion Attention(MSFFA)network model based on ResNet is proposed. The multi-scale attention module fuses different receptive fields for attention feature enhancement. Further, to reduce the loss of feature information fusion, a cross-scale feature fusion module is used to enhance the features of similar and edge domains. Finally, the feature information of the original sample is integrated into the segmentation results for detail enhancement, further optimizing the segmentation performance of the model. Experimental results show that compared with U-Net, AttUNet, and other models, the MSFFA model provides a more complete cervical vertebrae structure and smoother edges and can achieve more accurate segmentation results in lumbar vertebrae segmentation. Compared with the best model, DeepLabv3+, the mean Dice Similarity Coefficient(DSC)increases by 1.05 percentage points.

Key words: cervical vertebrae segmentation, attention mechanism, multi-scale fusion, feature enhancement, Convolutional Neural Network(CNN)

周静, 钟原, 李平, 杨毅, 马立泰, 张涛. 用于颈椎MRI分割的多尺度特征融合注意力网络模型[J]. 计算机工程, 2023, 49(10): 298-304.

Jing ZHOU, Yuan ZHONG, Ping LI, Yi YANG, Litai MA, Tao ZHANG. Multi-Scale Feature Fusion Attention Network Model for Cervical Vertebrae MRI Segmentation[J]. Computer Engineering, 2023, 49(10): 298-304.

http://www.ecice06.com/CN/Y2023/V49/I10/298

图/表 12

图1 MSFFA网络模型整体结构

Fig.1 Overall structure of MSFFA network model

图2 多尺度注意力结构

Fig.2 Structure of multi-scale attention

图3 跨尺度特征融合结构

Fig.3 Structure of cross-scale feature fusion

图4 不同模型在颈椎数据集上的分割结果

Fig.4 Segmentation results of different models on cervical vertebra dataset

图5 模块叠加的性能变化

Fig.5 Performance change of module overlays

图6 不同特征组合的性能差异

Fig.6 Performance difference of different feature combinations

图7 不同模型的平均训练时间和内存消耗

Fig.7 Average training time and memory consumption of different models

图8 不同模型在腰椎数据集上的分割结果

Fig.8 Segmentation results of different models on lumbar vertebra dataset

参考文献 37

1	CUNHA G M, FOWLER K J. Automated liver segmentation for quantitative MRI analysis. Radiology, 2022, 302 (2): 355- 356. doi: 10.1148/radiol.2021212306
2	TONG Q Q, LI C Z, SI W X, et al. RIANet: recurrent interleaved attention network for cardiac MRI segmentation. Computers in Biology and Medicine, 2019, 109, 290- 302. doi: 10.1016/j.compbiomed.2019.04.042
3	HUA L, GU Y, GU X Q, et al. A novel brain MRI image segmentation method using an improved multi-view fuzzy c-means clustering algorithm. Frontiers in Neuroscience, 2021, 15, 662674. doi: 10.3389/fnins.2021.662674
4	JIN Q G, MENG Z P, PHAM T D, et al. DUNet: a deformable network for retinal vessel segmentation. Knowledge-Based Systems, 2019, 178, 149- 162. doi: 10.1016/j.knosys.2019.04.025
5	HAMET P, TREMBLAY J. Artificial intelligence in medicine. Metabolism, 2017, 69, 36- 40. doi: 10.1016/j.metabol.2017.01.011
6	TOH T S, DONDELINGER F, WANG D. Looking beyond the hype: applied AI and machine learning in translational medicine. eBioMedicine, 2019, 47, 607- 615. doi: 10.1016/j.ebiom.2019.08.027
7	ZHANG L, WANG H. A novel segmentation method for cervical vertebrae based on PointNet++ and converge segmentation. Computer Methods and Programs in Biomedicine, 2021, 200, 105798. doi: 10.1016/j.cmpb.2020.105798
8	HILLE G, SAALFELD S, SEROWY S, et al. Vertebral body segmentation in wide range clinical routine spine MRI data. Computer Methods and Programs in Biomedicine, 2018, 155, 93- 99. doi: 10.1016/j.cmpb.2017.12.013
9	RAK M, STEFFEN J, MEYER A, et al. Combining convolutional neural networks and star convex cuts for fast whole spine vertebra segmentation in MRI. Computer Methods and Programs in Biomedicine, 2019, 177, 47- 56. doi: 10.1016/j.cmpb.2019.05.003
10	SHARMA P, SHUKLA A P. A review on brain tumor segmentation and classification for MRI images[C]//Proceedings of International Conference on Advance Computing and Innovative Technologies in Engineering. Washington D. C., USA: IEEE Press, 2021: 963-967.
11	RAO C S, KARUNAKARA K. A comprehensive review on brain tumor segmentation and classification of MRI images. Multimedia Tools and Applications, 2021, 80 (12): 17611- 17643. doi: 10.1007/s11042-020-10443-1
12	LU Y N, ZHAO Y, CHEN X, et al. A novel U-Net based deep learning method for 3D cardiovascular MRI segmentation[EB/OL]. [2022-08-17]. https://pubmed.ncbi.nlm.gov/35634040.
13	ZHUANG X H, SHEN J. Multi-scale patch and multi-modality atlases for whole heart segmentation of MRI. Medical Image Analysis, 2016, 31, 77- 87. doi: 10.1016/j.media.2016.02.006
14	KOLARIK M, BURGET R, RIHA K, et al. Suitability of CT and MRI imaging for automatic spine segmentation using deep learning[C]//Proceedings of the 44th International Conference on Telecommunications and Signal Processing. Washington D. C., USA: IEEE Press, 2021: 390-393.
15	VANIA M, MUREJA D, LEE D. Automatic spine segmentation from CT images using convolutional neural network via redundant generation of class labels. Journal of Computational Design and Engineering, 2019, 6 (2): 224- 232. doi: 10.1016/j.jcde.2018.05.002
16	LIU X X, DENG W X, LIU Y. Application of hybrid network of UNet and feature pyramid network in spine segmentation[C]//Proceedings of IEEE International Symposium on Medical Measurements and Applications. Washington D. C., USA: IEEE Press, 2021: 1-6.
17	PANG S M, PANG C L, SU Z H, et al. DGMSNet: spine segmentation for MR image by a detection-guided mixed-supervised segmentation network. Medical Image Analysis, 2022, 75, 102261. doi: 10.1016/j.media.2021.102261
18	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2016: 770-778.
19	CHEN L C, PAPANDREOU G, KOKKINOS I, et al. DeepLab: semantic image segmentation with deep convolutional Nets, Atrous convolution, and fully connected CRFs. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40 (4): 834- 848. doi: 10.1109/TPAMI.2017.2699184
20	景庄伟, 管海燕, 彭代峰, 等. 基于深度神经网络的图像语义分割研究综述. 计算机工程, 2020, 46 (10): 1- 17. URL
	JING Z W, GUAN H Y, PENG D F, et al. Survey of research in image semantic segmentation based on deep neural network. Computer Engineering, 2020, 46 (10): 1- 17. URL
21	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
22	吉彬, 任建君, 郑秀娟, 等. 改进U-Net在喉白斑病灶分割中的应用. 计算机工程, 2020, 46 (9): 248- 253. URL
	JI B, REN J J, ZHENG X J, et al. Application of improved U-Net in segmentation of laryngeal leukoplakia lesion. Computer Engineering, 2020, 46 (9): 248- 253. URL
23	LONG J, SHELHAMER E, DARRELL T. Fully convolutional networks for semantic segmentation[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2015: 3431-3440.
24	SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[EB/OL]. [2022-08-17]. https://arxiv.org/abs/1409.1556.
25	CHEN L C, PAPANDREOU G, KOKKINOS I, et al. Semantic image segmentation with deep convolutional Nets and fully connected CRFs[EB/OL]. [2022-08-17]. https://arxiv.org/abs/1412.7062.
26	CHEN L C, PAPANDREOU G, SCHROFF F, et al. Rethinking Atrous convolution for semantic image segmentation[EB/OL]. [2022-08-17]. https://arxiv.org/abs/1706.05587.
27	CHEN L C, ZHU Y K, PAPANDREOU G, et al. Encoder-decoder with Atrous separable convolution for semantic image segmentation[C]//Proceedings of European Conference on Computer Vision. Berlin, Germany: Springer, 2018: 801-818.
28	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.
29	XIAO X N, LIAN S, LUO Z, et al. Weighted Res-UNet for high-quality retina vessel segmentation[C]//Proceedings of the 9th International Conference on Information Technology in Medicine and Education. Washington D. C., USA: IEEE Press, 2018: 327-331.
30	LI X M, CHEN H, QI X J, et al. H-DenseUNet: hybrid densely connected UNet for liver and tumor segmentation from CT volumes. IEEE Transactions on Medical Imaging, 2018, 37 (12): 2663- 2674. doi: 10.1109/TMI.2018.2845918
31	ZHOU Z W, RAHMAN SIDDIQUEE M M, TAJBAKHSH N, et al. UNet++: a nested U-Net architecture for medical image segmentation. Berlin, Germany: Springer, 2018.
32	WANG G T, LI W Q, OURSELIN S, et al. Automatic brain tumor segmentation using convolutional neural networks with test-time augmentation[C]//Proceedings of International MICCAI Brainlesion Workshop. Berlin, Germany: Springer, 2019: 61-72.
33	OKTAY O, SCHLEMPER J, LE FOLGOC L, et al. Attention U-Net: learning where to look for the pancreas[EB/OL]. [2022-08-17]. https://arxiv.org/abs/1804.03999.
34	CHEN J N, LU Y Y, YU Q H, et al. TransUNet: transformers make strong encoders for medical image segmentation[EB/OL]. [2022-08-17]. https://arxiv.org/abs/2102.04306.
35	GAO Y H, ZHOU M, METAXAS D N. UTNet: a hybrid transformer architecture for medical image segmentation[C]//Proceedings of International Conference on Medical Image Computing and Computer Assisted Intervention. Berlin, Germany: Springer, 2021: 61-71.
36	CAO H, WANG Y Y, CHEN J, et al. Swin-UNet: UNet-like pure transformer for medical image segmentation[EB/OL]. [2022-08-17]. https://arxiv.org/abs/2105.05537.
37	MASOOD R F, HASSAN T, RAJA H, et al. A composite dataset of lumbar spine images with mid-sagittal view annotations and clinically significant spinal measurements[C]//Proceedings of the 2nd International Conference on Digital Futures and Transformative Technologies. Washington D. C., USA: IEEE Press, 2022: 1-5.

[1]	朱孟栩, 张文豪, 李国洪, 顾行发, 余涛, 郑逢杰, 张丽丽, 吴俣, 邴芳飞, 唐健雄. 基于卷积神经网络的高分六号卫星多光谱图像压缩[J]. 计算机工程, 2023, 49(9): 287-294.
[2]	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.
[3]	李现国, 李滨. 基于Transformer和多尺度CNN的图像去模糊[J]. 计算机工程, 2023, 49(9): 226-233, 245.
[4]	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.
[5]	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.
[6]	杜逸潇, 王红军, 李修和. 基于频谱地图的辐射源指纹定位方法研究[J]. 计算机工程, 2023, 49(9): 183-190, 198.
[7]	胡水. 基于深度强化学习的智能兵棋推演决策方法[J]. 计算机工程, 2023, 49(9): 303-312.
[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]	李哲铭, 王晋东, 侯建中, 李伟, 张世华, 张恒巍. 基于显著区域优化的对抗样本攻击方法[J]. 计算机工程, 2023, 49(9): 246-255, 264.
[10]	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.
[11]	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.
[12]	卢昂, 储珺, 冷璐. 基于高低频特征增强的图像去雾[J]. 计算机工程, 2023, 49(8): 174-181.
[13]	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.
[14]	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.
[15]	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.

选择文件类型/文献管理软件名称

选择包含的内容