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RAU-net的视网膜OCT图像快速自动分层研究

曾兴晖 许祥丛 李晓 王茗祎 钟俊平 熊红莲

曾兴晖, 许祥丛, 李晓, 王茗祎, 钟俊平, 熊红莲. 基于RAU-net的视网膜OCT图像快速自动分层研究[J]. 华南师范大学学报(自然科学版), 2021, 53(2): 1-6. doi: 10.6054/j.jscnun.2021019
引用本文: 曾兴晖, 许祥丛, 李晓, 王茗祎, 钟俊平, 熊红莲. 基于RAU-net的视网膜OCT图像快速自动分层研究[J]. 华南师范大学学报(自然科学版), 2021, 53(2): 1-6. doi: 10.6054/j.jscnun.2021019
ZENG Xinghui, XU Xiangcong, LI Xiao, WANG Mingyi, ZHONG Junping, XIONG Honglian. Quick and Automatic Segmentation of Retinal Layers in OCT Images Using Residual and Attention U-net[J]. Journal of South China normal University (Natural Science Edition), 2021, 53(2): 1-6. doi: 10.6054/j.jscnun.2021019
Citation: ZENG Xinghui, XU Xiangcong, LI Xiao, WANG Mingyi, ZHONG Junping, XIONG Honglian. Quick and Automatic Segmentation of Retinal Layers in OCT Images Using Residual and Attention U-net[J]. Journal of South China normal University (Natural Science Edition), 2021, 53(2): 1-6. doi: 10.6054/j.jscnun.2021019

基于RAU-net的视网膜OCT图像快速自动分层研究

doi: 10.6054/j.jscnun.2021019
基金项目: 

国家自然科学基金项目 81601534

国家自然科学基金项目 61705036

广东省自然科学基金项目 2017A030313386

详细信息
    通讯作者:

    熊红莲, yuanxiufeng138@163.com

  • 中图分类号: TP751

Quick and Automatic Segmentation of Retinal Layers in OCT Images Using Residual and Attention U-net

  • 摘要: 利用计算机图像处理技术自动分割视网膜图像,获得各层的厚度,可对多种视网膜疾病进行直观评估. 为了快速准确地对OCT视网膜图像进行自动分层,提出一种结合RAU-net和图像搜索的视网膜图像自动分层方法. 该方法在U-net的基础上加入了残差块结构和注意力门结构,残差块结构在构建更深的网络、获取高级特征的同时,有效避免了梯度消失和梯度爆炸问题,注意力门结构突出了模型对视网膜图像重要特征的学习. 由模型预测后得到的分割结果获取9条边界的感兴趣区域,然后运用图像搜索对分层图像进行边界优化,得到精确的视网膜分层图像. 结果表明:该RAU-net算法与手动分层的误差约为1像素,且完成1帧OCT视网膜图像的分层只需要4 s. 通过RAU-net与图像搜索相结合的方法为视网膜疾病的临床诊断和治疗提供了快速准确的定量分析方法.
  • 图  1  算法流程图

    Figure  1.  The flowchart of algorithm

    图  2  残差块结构

    Figure  2.  The structure of residual block

    图  3  注意力门的结构

    Figure  3.  The attention gate structure

    图  4  RAU-net结构

    Figure  4.  The structure of RAU-net

    图  5  节点图分割过程

    Figure  5.  The process of node graph segmentation

    图  6  准确率和损失函数曲线

    Figure  6.  The accuracy and loss curves

    图  7  视网膜分割

    Figure  7.  The retina segmentation

    图  8  9条边界的ROI区域

    Figure  8.  The ROI area of nine borders

    图  9  本文与FANG的方法[11]分割的结果

    Figure  9.  The segmentation results of the present method and FANG's method[11]

    表  1  2种方法的像素级评估

    Table  1.   The pixel-level evaluation of both methods

    边界 MAE RMSE
    FANG的方法[11] RAU-net+Dijkstra FANG的方法[11] RAU-net+Dijkstra
    ILM 1.686 2 0.899 4 2.101 7 1.222 0
    RNFL-GCL 1.661 7 1.485 4 2.630 0 1.936 7
    IPL-INL 3.395 7 1.656 3 4.163 2 2.127 3
    INL-OPL 2.815 8 0.798 8 3.760 8 1.111 0
    OPL-ONL 1.652 3 1.219 7 2.063 0 1.723 9
    ELM 2.482 7 0.707 0 3.219 1 1.084 3
    IS-OS 1.243 6 0.693 4 1.593 2 0.960 1
    OS-RPE 1.458 1 0.986 3 1.889 1 1.273 2
    BM 1.701 5 1.138 7 2.080 6 1.519 4
    注:受试数为5个.
    下载: 导出CSV
  • [1] HUANG D, SWANSON E A, LIN C P, et al. Optical coherence tomography[J]. Science, 1991, 254: 1178-1181. doi: 10.1126/science.1957169
    [2] 朱良慧, 曾毛毛, 赵佳玮, 等. 基于OCT图像的组织散射系数提取方法及其应用[J]. 华南师范大学学报(自然科学版), 2016, 48(4): 31-34. doi: 10.6054/j.jscnun.2016.05.008

    ZHU L H, ZENG M M, ZHAO J W, et al. Quantify wound skin by extracting tissue scattering coefficient from OCT image[J]. Joumal of South China Normal University (Natural Science Edition), 2016, 48(4): 31-34. doi: 10.6054/j.jscnun.2016.05.008
    [3] READ S A, ALONSO-CANEIRO D, VINCENT S J. Longitudinal changes in macular retinal layer thickness in pediatric populations: myopic vs non-myopic eyes[J]. PloS One, 2017, 12(6): e0180462/1-22. http://pubmedcentralcanada.ca/pmcc/articles/PMC5491256/
    [4] CHAUHAN B C, VIANNA J R, SHARPE G P, et al. Differential effects of aging in the macular retinal layers, neuroretinal rim, and peripapillary retinal nerve fiber layer[J]. Ophthalmology, 2020, 127(2): 177-185. doi: 10.1016/j.ophtha.2019.09.013
    [5] BUSSEL I I, WOLLSTEIN G, SCHUMAN J S. OCT for glaucoma diagnosis, screening and detection of glaucoma progression[J]. British Journal of Ophthalmology, 2014, 98(S2): 15-19.
    [6] LEE W J, KIM Y K, KIM Y W, et al. Rate of macular ganglion cell-inner plexiform layer thinning in glaucomatous eyes with vascular endothelial growth factor inhibition[J]. Journal of Glaucoma, 2017, 26(11): 980-986. doi: 10.1097/IJG.0000000000000776
    [7] NOVOSEL J, THEPASS G, LEMIJ H G, et al. Loosely coupled level sets for simultaneous 3D retinal layer segmentation in optical coherence tomography[J]. Medical Image Analysis, 2015, 26(1): 146-158. doi: 10.1016/j.media.2015.08.008
    [8] SUN Y, NIU S, GAO X, et al. Adaptive-guided-coupling-probability level set for retinal layer segmentation[J]. IEEE Journal of Biomedical and Health Informatics, 2020, 24(11): 3236-3247. doi: 10.1109/JBHI.2020.2981562
    [9] GARVIM M K, ABRAMOFF M D, WU X, et al. Automated 3-D intraretinal layer segmentation of macular spectral-domain optical coherence tomography images[J]. IEEE Transactions on Medical Imaging, 2009, 28(9): 1436-1447. doi: 10.1109/TMI.2009.2016958
    [10] CHIU S J, LI X T, NICHOLAS P, et al. Automatic segmentation of seven retinal layers in SDOCT images congruent with expert manual segmentation[J]. Optics Express, 2010, 18: 19413-19428. doi: 10.1364/OE.18.019413
    [11] FANG L, CUNEFARE D, WANG C, et al. Automatic segmentation of nine retinal layer boundaries in OCT images of non-exudative AMD patients using deep learning and graph search[J]. Biomedical Optics Express, 2017, 8(5): 2732-2744. doi: 10.1364/BOE.8.002732
    [12] LANG A, CARASS A, HAUSER M, et al. Retinal layer segmentation of macular OCT images using boundary classification[J]. Biomedical Optics Express, 2013, 4(7): 1133-1152. doi: 10.1364/BOE.4.001133
    [13] LIU Y, CARASS A, SOLOMON S D, et al. Multi-layer fast level set segmentation for macular OCT[C]//Proceedings of the 2018 IEEE 15th International Symposium on Biomedical Imaging. Washington: IEEE, 2018: 1445-1448.
    [14] 唐小煜, 黄进波, 冯洁文, 等. 基于U-net和YOLOv4的绝缘子图像分割与缺陷检测[J]. 华南师范大学学报(自然科学版), 2020, 52(6): 15-21. doi: 10.6054/j.jscnun.2020088

    TANG X Y, HUANG J B, FENG J W, et al. Image segmentation and defect detection of insulators based on U-net and YOLOv4[J]. Joumal of South China Normal University (Natural Science Edition), 2020, 52(6): 15-21. doi: 10.6054/j.jscnun.2020088
    [15] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. Las Vegas: IEEE, 2016: 770-778.
    [16] DIJKSTRA E W. A note on two problems in connexion with graphs[J]. Numerische Mathematik, 1959, 1(1): 269-271. doi: 10.1007/BF01386390
    [17] HINTON G E, SRIVASTAVA N, KRIZHEVSKY A, et al. Improving neural networks by preventing co-adaptation of feature detectors[J/OL]. (2012-07-03)[2020-06-18]. https://arxiv.org/pdf/1207.0580.
    [18] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//Proceedings of neural information processing systems. Long Beach: NIPS, 2017: 5998-6008.
    [19] OKTAY O, SCHLEMPER J, FOLGOC L L, et al. Attention U-net: learning where to look for the pancreas[J/OL]. (2018-05-20)[2020-06-18]. https://arxiv.org/pdf/1804.03999.
  • 加载中
图(9) / 表(1)
计量
  • 文章访问数:  152
  • HTML全文浏览量:  75
  • PDF下载量:  25
  • 被引次数: 0
出版历程
  • 收稿日期:  2020-06-18
  • 网络出版日期:  2021-04-29
  • 刊出日期:  2021-04-25

目录

    /

    返回文章
    返回