留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于特征图金字塔的冠脉造影图像血管分割方法

郭昊虎 高若谦 葛明锋 董文飞 刘炎 赵旭峰

郭昊虎, 高若谦, 葛明锋, 董文飞, 刘炎, 赵旭峰. 基于特征图金字塔的冠脉造影图像血管分割方法[J]. 中国光学(中英文). doi: 10.37188/CO.2023-0186
引用本文: 郭昊虎, 高若谦, 葛明锋, 董文飞, 刘炎, 赵旭峰. 基于特征图金字塔的冠脉造影图像血管分割方法[J]. 中国光学(中英文). doi: 10.37188/CO.2023-0186
GUO Hao-hu, GAO Ruo-qian, GE Ming-feng, DONG Wen-fei, LIU Yan, ZHAO Xu-feng. Coronary artery angiography image vessel segmentation method based on Feature Pyramid Network[J]. Chinese Optics. doi: 10.37188/CO.2023-0186
Citation: GUO Hao-hu, GAO Ruo-qian, GE Ming-feng, DONG Wen-fei, LIU Yan, ZHAO Xu-feng. Coronary artery angiography image vessel segmentation method based on Feature Pyramid Network[J]. Chinese Optics. doi: 10.37188/CO.2023-0186

基于特征图金字塔的冠脉造影图像血管分割方法

doi: 10.37188/CO.2023-0186
基金项目: 国家重点研发计划(No. 2021YFC2500500);吉林省与中国科学院科技合作高新技术产业化专项资金项目(No. 2023SYHZ0037)
详细信息
    作者简介:

    郭昊虎(1998—),男,吉林长春人,硕士研究生,2020 年于长春工业大学获得学士学位,主要从事图像处理、深度学习等方面的研究。E-mail:441804766@qq.com

    高若谦(1993—),男,吉林长春人,博士,博士后,主要从事高光谱、成像光学等方面研究。E-mail: gaorq@sibet.ac.cn

  • 中图分类号: TP391.41;

Coronary artery angiography image vessel segmentation method based on Feature Pyramid Network

Funds: Supported by the National Key R&D Program of China (No. 2021YFC2500500); Jilin Provincial Academy Science and Technology Cooperation Special Project (No. 2023SYHZ0037)
More Information
  • 摘要:

    针对冠脉造影图像照明不均,血管结构与背景区域对比度低,冠脉血管拓扑结构复杂等分割难点,建立了一个冠脉造影血管分割标注数据集,提出了一种基于特征图金字塔的冠脉造影图像血管分割模型。本文模型以U-Net网络为基础进行改进和优化,首先,将U-Net编码部分的第一个卷积层修改为一个7×7的卷积层,提高每一层的感受野,在编解码层中引入修改后的ConvNeXt block,提升网络提取更深层次特征的能力;其次,设计分组注意力机制模块GA引入到U-Net跨连接处,对编码部分提取的特征进行增强,弥补编解码器间的存在语义差距;最后,在U-Net解码器处设计了一个特征图金字塔级联模块PFC,融合各尺度的特征图,并在PFC中每一层加入SE注意力机制模块筛选特征图中有效信息,网络的损失函数为PFC模块各层输出的加权,以监督网络各层的特征提取。本文模型在测试集上的测试结果如下:Dice系数为0.8843,Jaccard系数为0.7926。实验结果表明,相比其他对比方法,本文模型在冠脉血管分割上具有较强的鲁棒性,在低对比度下能够有效抑制噪声,对冠脉血管具有更好的分割效果。

     

  • 图 1  模型网络结构

    Figure 1.  Model network structure

    图 2  ConvNeXt模块

    Figure 2.  ConvNeXt block

    图 3  修改后的ConvNeXt模块

    Figure 3.  Modified ConvNeXt block

    图 4  特征金字塔网络

    Figure 4.  Feature Pyramid Network

    图 5  金字塔特征级联(PFC)

    Figure 5.  Pyramid Feature Concatenation(PFC)

    图 6  SE注意力机制

    Figure 6.  SE attention mechanism

    图 7  分组注意力机制模块

    Figure 7.  Group Attention block

    图 8  验证集上不同算法的DICE曲线

    Figure 8.  Dice curves of different algorithms on the testing set

    图 9  不同算法分割效果对比图

    Figure 9.  Comparison chart of different algorithm segmentation effects

    表  1  各模块对性能的影响

    Table  1.   Each module’s impact on performance

    网络JaccardDice
    BaseNet0.6476±0.00980.7861±0.0073
    BaseNet+Conv.7×70.6606±0.01190.7956±0.0086
    BaseNet+ConvNeXt block0.7104±0.00650.8307±0.0044
    BaseNet+修改后的 ConvNeXt block0.7234±0.00440.8395±0.0030
    BaseNet+PFC0.7036±0.00880.8260±0.0061
    BaseNet+PFC+SE0.7104±0.00430.8260±0.0061
    BaseNet+PFC+SE+加权Loss0.7364±0.00620.8482±0.0041
    BaseNet+GA0.7044±0.00280.8266±0.0019
    BaseNet+Conv.7×7 +PFC+SE +修改后的ConvNeXt block+GA+加权Loss0.7926±0.00580.8843±0.0036
    下载: 导出CSV

    表  2  不同算法测试结果

    Table  2.   Test results for different algorithms

    网络JaccardDiceAUCAccuracyPrecisionSensitivitySpecificity
    U-Net0.71160.83150.97330.97060.87690.79080.9888
    ResUNeXt0.68490.81300.87410.96830.88470.75260.9901
    TransUnet0.74210.85190.98840.97480.86280.84160.9873
    MultiResUnet0.76640.86770.98420.97610.87750.85830.9879
    Ours0.79260.88430.99130.97830.90080.85970.9904
    下载: 导出CSV
  • [1] WANG H, NAGHAVI M, ALLEN C, et al. Global, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for 249 causes of death, 1980-2015: a systematic analysis for the Global Burden of Disease Study 2015[J]. The Lancet, 2016, 388(10053): 1459-1544.
    [2] BAIM D S, GROSSMAN W. Coronary angiography. 1986. (查阅网上资料, 未找到本条文献信息, 请确认) .
    [3] FLEMING R M, KIRKEEIDE R L, SMALLING R W, et al. Patterns in visual interpretation of coronary arteriograms as detected by quantitative coronary arteriography[J]. Journal of the American College of Cardiology, 1991, 18(4): 945-951.
    [4] WIJNS W, SERRUYS P W, REIBER J H, et al. Quantitative angiography of the left anterior descending coronary artery: correlations with pressure gradient and results of exercise thallium scintigraphy[J]. Circulation, 1985, 71(2): 273-279.
    [5] GARRONE P, BIONDI-ZOCCAI G, SALVETTI I, et al. Quantitative coronary angiography in the current era: principles and applications[J]. Journal of Interventional Cardiology, 2009, 22(6): 527-536.
    [6] BLONDEL C, MALANDAIN G, VAILLANT R, et al. Reconstruction of coronary arteries from a single rotational X-ray projection sequence[J]. IEEE Transactions on Medical Imaging, 2006, 25(5): 653-663.
    [7] SHECHTER G, DEVERNAY F, COSTE-MANIÈRE E, et al. Three-dimensional motion tracking of coronary arteries in biplane cineangiograms[J]. IEEE Transactions on Medical Imaging, 2003, 22(4): 493-503.
    [8] SUN ZH, ZHOU Y. Assessing cardiac dynamics based on X-ray coronary angiograms[J]. J. Multim., 2013, 8(1): 48-55.
    [9] FELFELIAN B, FAZLALI H R, KARIMI N, et al. Vessel segmentation in low contrast X-ray angiogram images[C]. 2016 IEEE International Conference on Image Processing (ICIP), IEEE, 2016: 375-379.
    [10] CHEN Y, ZHANG Y D, YANG J, et al. Curve-like structure extraction using minimal path propagation with backtracking[J]. IEEE Transactions on Image Processing, 2016, 25(2): 988-1003.
    [11] JIN M X, LI R, JIANG J, et al. Extracting contrast-filled vessels in X-ray angiography by graduated RPCA with motion coherency constraint[J]. Pattern Recognition, 2017, 63: 653-666.
    [12] BANKHEAD P, SCHOLFIELD C N, MCGEOWN J G, et al. Fast retinal vessel detection and measurement using wavelets and edge location refinement[J]. PLoS One, 2012, 7(3): e32435.
    [13] LI Y L, ZHOU SH J, WU J H, et al. A novel method of vessel segmentation for X-ray coronary angiography images[C]. 2012 Fourth International Conference on Computational and Information Sciences, IEEE, 2012: 468-471.
    [14] SOARES J V B, LEANDRO J J G, CESAR R M, et al. Retinal vessel segmentation using the 2-D gabor wavelet and supervised classification[J]. IEEE Transactions on Medical Imaging, 2006, 25(9): 1214-1222.
    [15] FRANGI A F, NIESSEN W J, VINCKEN K L, et al. Multiscale vessel enhancement filtering[C]. Proceedings of the 1st International Conference on Medical Image Computing and Computer-Assisted Intervention — MICCAI’98, Springer, 1998: 130-137.
    [16] M’HIRI F, DUONG L, DESROSIERS C, et al. Vessel walker: coronary arteries segmentation using random walks and hessian-based vesselness filter[C]. 2013 IEEE 10th International Symposium on Biomedical Imaging, IEEE, 2013: 918-921.
    [17] DEHKORDI M T, HOSEINI A M D, SADRI S, et al. Local feature fitting active contour for segmenting vessels in angiograms[J]. IET Computer Vision, 2014, 8(3): 161-170.
    [18] LAW M W K, CHUNG A C S. Efficient implementation for spherical flux computation and its application to vascular segmentation[J]. IEEE Transactions on Image Processing, 2009, 18(3): 596-612.
    [19] ORLANDO J I, PROKOFYEVA E, BLASCHKO M B. A discriminatively trained fully connected conditional random field model for blood vessel segmentation in fundus images[J]. IEEE Transactions on Bio-Medical Engineering, 2017, 64(1): 16-27.
    [20] 郑跃坤, 葛明锋, 常智敏, 等. 基于残差网络的结直肠内窥镜图像超分辨率重建方法[J]. 中国光学(中英文),2023,16(5):1022-1033.

    ZHENG Y K, GE M F, CHANG ZH M, et al. Super-resolution reconstruction for colorectal endoscopic images based on a residual network[J]. Chinese Optics, 2023, 16(5): 1022-1033. (in Chinese).
    [21] 白瑞峰, 江山, 孙海江, 等. 基于编码解码结构的微血管减压图像实时语义分割[J]. 中国光学(中英文),2022,15(5):1055-1065.

    BAI R F, JIANG SH, SUN H J, et al. Real-time semantic segmentation of microvascular decompression images based on encoder-decoder structure[J]. Chinese Optics, 2022, 15(5): 1055-1065. (in Chinese).
    [22] ZHAO H SH, SHI J P, QI X J, et al. Pyramid scene parsing network[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2017: 6230-6239.
    [23] BADRINARAYANAN V, KENDALL A, CIPOLLA R. SegNet: a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12): 2481-2495.
    [24] RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]. Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015, Springer, 2015: 234-241.
    [25] IBTEHAZ N, RAHMAN M S. MultiResUNet: rethinking the U-Net architecture for multimodal biomedical image segmentation[J]. Neural Networks, 2020, 121: 74-87.
    [26] PAN S W, ZHANG W, ZHANG W J, et al. Diagnostic model of coronary microvascular disease combined with full convolution deep network with balanced cross-entropy cost function[J]. IEEE Access, 2019, 7: 177997-178006.
    [27] XIAN ZH CH, WANG X Q, YAN SH D, et al. Main coronary vessel segmentation using deep learning in smart medical[J]. Mathematical Problems in Engineering, 2020, 2020: 8858344.
    [28] YANG S, KWEON J, KIM Y H. Major vessel segmentation on X-ray coronary angiography using deep networks with a novel penalty loss function[C]. Proceedings of Machine Learning Research, MIDL, 2019.
    [29] JUN T J, KWEON J, KIM Y H, et al. T-Net: Nested encoder-decoder architecture for the main vessel segmentation in coronary angiography[J]. Neural Networks, 2020, 128: 216-233.
    [30] LI L ZH, VERMA M, NAKASHIMA Y, et al. IterNet: retinal image segmentation utilizing structural redundancy in vessel networks[C]. 2020 IEEE Winter Conference on Applications of Computer Vision (WACV), IEEE, 2020: 3645-3654.
    [31] LIU ZH, MAO H Z, WU CH Y, et al. A ConvNet for the 2020s[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE, 2022: 11966-11976.
    [32] LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2017: 936-944.
    [33] HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE, 2018: 7132-7141.
    [34] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]. Proceedings of the 31st International Conference on Neural Information Processing Systems, NIPS, 2017: 5998-6008.
    [35] DIAKOGIANNIS F I, WALDNER F, CACCETTA P, et al. ResUNet-a: a deep learning framework for semantic segmentation of remotely sensed data[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2020, 162: 94-114.
    [36] CHEN J N, LU Y Y, YU Q H, et al. TransUNet: transformers make strong encoders for medical image segmentation[J]. arXiv: 2102.04306, 2021. (查阅网上资料, 不确定文献类型及格式是否正确, 请确认) .
  • 加载中
图(9) / 表(2)
计量
  • 文章访问数:  73
  • HTML全文浏览量:  15
  • PDF下载量:  9
  • 被引次数: 0
出版历程
  • 网络出版日期:  2024-05-09

目录

    /

    返回文章
    返回

    重要通知

    2024年2月16日科睿唯安通过Blog宣布,2024年将要发布的JCR2023中,229个自然科学和社会科学学科将SCI/SSCI和ESCI期刊一起进行排名!《中国光学(中英文)》作为ESCI期刊将与全球SCI期刊共同排名!