计算机科学与探索 ›› 2022, Vol. 16 ›› Issue (8): 1877-1884.DOI: 10.3778/j.issn.1673-9418.2012011
杨知桥1, 张莹1,+(), 王新杰1, 张东波1,2, 王玉1
收稿日期:
2020-12-03
修回日期:
2021-01-28
出版日期:
2022-08-01
发布日期:
2021-02-05
通讯作者:
+E-mail: zhangying@xtu.edu.cn。作者简介:
杨知桥(1996—),男,湖南桃源人,硕士研究生,主要研究方向为医学图像处理。基金资助:
YANG Zhiqiao1, ZHANG Ying1,+(), WANG Xinjie1, ZHANG Dongbo1,2, WANG Yu1
Received:
2020-12-03
Revised:
2021-01-28
Online:
2022-08-01
Published:
2021-02-05
About author:
YANG Zhiqiao, born in 1996, M.S. candidate. His research interest is medical image processing.Supported by:
摘要:
眼底视网膜血管分析和渗出物、出血点等主要病灶区检测是判断糖尿病性视网膜病变程度的重要方法。针对细微血管的分叉以及端点处分割效果不好、渗出物边界不明显以及出血点细小且分布零散不易分割等问题,提出一种改进U型网络,通过改进上下文提取编码模块,提取更丰富的高级别特征;并在特征编码阶段加入混合注意力机制(HAM),突出细微血管以及病灶区特征,减小背景类和噪声影响。实验结果表明,提出的算法在眼底视网膜血管分割数据集DRIVE上的分割准确率、灵敏度、特异性和AUC值比U-NET、CE-NET等现有方法有一定提升,其中灵敏度相较CE-Net网络提升了0.014 6。在糖尿病性视网膜病变病灶区分割数据集DIARETDB1上,对渗出物和出血点的分割效果比U-NET、CE-NET等现有方法有较好的提升,能有效辅助医生诊断。
中图分类号:
杨知桥, 张莹, 王新杰, 张东波, 王玉. 改进U型网络在视网膜病变检测中的应用研究[J]. 计算机科学与探索, 2022, 16(8): 1877-1884.
YANG Zhiqiao, ZHANG Ying, WANG Xinjie, ZHANG Dongbo, WANG Yu. Application Research of Improved U-shaped Network in Detection of Retinopathy[J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(8): 1877-1884.
Method | AC | SE | SP | AUC |
---|---|---|---|---|
U-Net[ | 0.953 1 | 0.753 7 | 0.970 4 | 0.960 1 |
ResU-Net[ | 0.952 7 | 0.756 9 | 0.981 6 | 0.974 3 |
Dense U-Net[ | 0.953 2 | 0.797 7 | 0.975 9 | 0.973 5 |
CE-Net(Backbone)[ | 0.954 5 | 0.830 9 | 0.979 8 | 0.977 9 |
Backbone+New DAC | 0.955 1 | 0.832 4 | 0.980 1 | 0.978 2 |
Backbone+New RMP | 0.956 3 | 0.833 2 | 0.980 6 | 0.978 7 |
Backbone+New DAC+ New RMP | 0.957 2 | 0.834 6 | 0.981 0 | 0.979 1 |
Backbone+HAM | 0.958 5 | 0.838 9 | 0.981 9 | 0.980 5 |
Proposed method | 0.961 3 | 0.845 5 | 0.984 3 | 0.982 7 |
表1 不同方法分割性能对比(眼底视网膜血管)
Table 1 Comparison of segmentation performance of different methods (fundus retinal blood vessels)
Method | AC | SE | SP | AUC |
---|---|---|---|---|
U-Net[ | 0.953 1 | 0.753 7 | 0.970 4 | 0.960 1 |
ResU-Net[ | 0.952 7 | 0.756 9 | 0.981 6 | 0.974 3 |
Dense U-Net[ | 0.953 2 | 0.797 7 | 0.975 9 | 0.973 5 |
CE-Net(Backbone)[ | 0.954 5 | 0.830 9 | 0.979 8 | 0.977 9 |
Backbone+New DAC | 0.955 1 | 0.832 4 | 0.980 1 | 0.978 2 |
Backbone+New RMP | 0.956 3 | 0.833 2 | 0.980 6 | 0.978 7 |
Backbone+New DAC+ New RMP | 0.957 2 | 0.834 6 | 0.981 0 | 0.979 1 |
Backbone+HAM | 0.958 5 | 0.838 9 | 0.981 9 | 0.980 5 |
Proposed method | 0.961 3 | 0.845 5 | 0.984 3 | 0.982 7 |
Method | AC | SE | SP | AUC |
---|---|---|---|---|
U-Net[ | 0.908 9 | 0.744 7 | 0.932 5 | 0.918 7 |
ResU-Net[ | 0.9177 | 0.812 6 | 0.939 8 | 0.920 2 |
Dense U-Net[ | 0.916 3 | 0.797 5 | 0.946 5 | 0.924 3 |
CE-Net(Backbone)[ | 0.917 5 | 0.826 5 | 0.944 6 | 0.929 8 |
Backbone+New DAC | 0.917 9 | 0.827 5 | 0.944 9 | 0.930 1 |
Backbone+New RMP | 0.918 4 | 0.828 6 | 0.945 3 | 0.930 3 |
Backbone+New DAC+ New RMP | 0.919 3 | 0.829 1 | 0.945 5 | 0.930 7 |
Backbone+HAM | 0.921 2 | 0.830 7 | 0.945 9 | 0.931 8 |
Proposed method | 0.923 7 | 0.833 2 | 0.947 3 | 0.933 5 |
表2 不同方法分割性能对比(渗出物)
Table 2 Comparison of segmentation performance of different methods (exudate)
Method | AC | SE | SP | AUC |
---|---|---|---|---|
U-Net[ | 0.908 9 | 0.744 7 | 0.932 5 | 0.918 7 |
ResU-Net[ | 0.9177 | 0.812 6 | 0.939 8 | 0.920 2 |
Dense U-Net[ | 0.916 3 | 0.797 5 | 0.946 5 | 0.924 3 |
CE-Net(Backbone)[ | 0.917 5 | 0.826 5 | 0.944 6 | 0.929 8 |
Backbone+New DAC | 0.917 9 | 0.827 5 | 0.944 9 | 0.930 1 |
Backbone+New RMP | 0.918 4 | 0.828 6 | 0.945 3 | 0.930 3 |
Backbone+New DAC+ New RMP | 0.919 3 | 0.829 1 | 0.945 5 | 0.930 7 |
Backbone+HAM | 0.921 2 | 0.830 7 | 0.945 9 | 0.931 8 |
Proposed method | 0.923 7 | 0.833 2 | 0.947 3 | 0.933 5 |
Method | AC | SE | SP | AUC |
---|---|---|---|---|
U-Net[ | 0.913 4 | 0.746 5 | 0.933 2 | 0.932 5 |
ResU-Net[ | 0.919 8 | 0.820 7 | 0.940 7 | 0.945 5 |
Dense U-Net[ | 0.916 8 | 0.794 4 | 0.944 3 | 0.943 9 |
CE-Net(Backbone)[ | 0.922 3 | 0.834 9 | 0.948 5 | 0.945 3 |
Backbone+New DAC | 0.923 3 | 0.835 5 | 0.948 9 | 0.945 8 |
Backbone+New RMP | 0.923 9 | 0.835 9 | 0.949 1 | 0.946 1 |
Backbone+New DAC+New RMP | 0.924 6 | 0.836 5 | 0.949 6 | 0.946 8 |
Backbone+HAM | 0.927 3 | 0.838 1 | 0.952 2 | 0.948 1 |
Proposed method | 0.931 0 | 0.839 8 | 0.955 4 | 0.949 6 |
表3 不同方法分割性能对比(出血点)
Table 3 Comparison of segmentation performance of different methods (bleeding point)
Method | AC | SE | SP | AUC |
---|---|---|---|---|
U-Net[ | 0.913 4 | 0.746 5 | 0.933 2 | 0.932 5 |
ResU-Net[ | 0.919 8 | 0.820 7 | 0.940 7 | 0.945 5 |
Dense U-Net[ | 0.916 8 | 0.794 4 | 0.944 3 | 0.943 9 |
CE-Net(Backbone)[ | 0.922 3 | 0.834 9 | 0.948 5 | 0.945 3 |
Backbone+New DAC | 0.923 3 | 0.835 5 | 0.948 9 | 0.945 8 |
Backbone+New RMP | 0.923 9 | 0.835 9 | 0.949 1 | 0.946 1 |
Backbone+New DAC+New RMP | 0.924 6 | 0.836 5 | 0.949 6 | 0.946 8 |
Backbone+HAM | 0.927 3 | 0.838 1 | 0.952 2 | 0.948 1 |
Proposed method | 0.931 0 | 0.839 8 | 0.955 4 | 0.949 6 |
[1] | CAVAN D, MAKAROFF L, DA ROCHA FERNANDES J, et al. The diabetic retinopathy barometer study: global pers-pectives on access to and experiences of diabetic retinopa-thy screening and treatment[J]. Diabetes Research and Cli-nical Practice, 2017, 129: 16-24. |
[2] | 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 Biomedical Engineering, 2016, 64(1): 16-27. |
[3] | MURUGESWARI S, SUKANESH R. Investigations of seve-rity level measurements for diabetic macular oedema using machine learning algorithms[J]. Irish Journal of Medical Science, 2017, 186(4): 929-938. |
[4] | QUELLEC G, CHARRIÈRE K, BOUDI Y, et al. Deep image mining for diabetic retinopathy screening[J]. Medical Image Analysis, 2017, 39: 178-193. |
[5] | RONNEBERGER O, FISCHER P, BROX T, et al. U-Net:convolutional networks for biomedical image segmentation[C]// LNCS 9351: Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention, Munich, Oct 5-9, 2015. Cham: Sprin-ger, 2015: 234-241. |
[6] | 钟思华, 郭兴明, 郑伊能. 改进U-Net网络的肺结节分割方法[J]. 计算机工程与应用, 2020, 56(17): 203-209. |
ZHONG S H, GUO X M, ZHENG Y N. Improved U-Net network for lung nodule segmentation[J]. Computer Engi-neering and Applications, 2020, 56(17): 203-209. | |
[7] | 钱宝鑫, 肖志勇, 宋威. 改进的卷积神经网络在肺部图像上的分割应用[J]. 计算机科学与探索, 2020, 14(8): 1358-1367. |
QIAN B X, XIAO Z Y, SONG W. Application of improved convolutional neural network in lung image segmentation[J]. Journal of Frontiers of Computer Science and Techno-logy, 2020, 14(8): 1358-1367. | |
[8] | GU Z, CHENG J, FU H, et al. CE-Net: context encoder net-work for 2D medical image segmentation[J]. IEEE Transac-tions on Medical Imaging, 2019, 38(10): 2281-2292. |
[9] | PANG S C, DU A A, YU Z M, et al. Correlation matters:multi-scale fine-grained contextual information extraction for hepatic tumor segmentation[C]// LNCS 12084: Procee-dings of the 24th Pacific-Asia Conference on Knowledge Discovery and Data Mining, Singapore, May 11-14, 2020. Cham: Springer, 2020: 462-474. |
[10] | WANG L, LIANG D X, YIN X L, et al. Coronary artery segmentation in angiographic videos utilizing spatial-tem-poral information[J]. BMC Medical Imaging, 2020, 20(1): 110. |
[11] | WANG P Q, CHEN P F, YUAN Y, et al. Understanding convolution for semantic segmentation[C]// Proceedings of the 2018 IEEE Winter Conference on Applications of Com-puter Vision, Lake Tahoe, Mar 12-15, 2018. Washington: IEEE Computer Society, 2018: 1451-1460. |
[12] | ZHAO H S, SHI J P, QI X J, et al. Pyramid scene parsing net-work[C]// Proceedings of the 2017 IEEE Conference on Com-puter Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Washington: IEEE Computer Society, 2017: 6230-6239. |
[13] | HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, Jun 18-22, 2018. Washington: IEEE Computer Society, 2018: 7132-7141. |
[14] | WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]// LNCS 11211: Proceedings of the 15th European Conference on Computer Vision, Munich, Sep 8-14, 2018. Cham: Springer, 2018: 3-19. |
[15] | CHEN L C, PAPANDREOU G, KOKKINOS I, et al. Deep-lab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 40(4): 834-848. |
[16] | STAAL J, ABRÀMOFF M D, NIEMEIJER M, et al. Ridge-based vessel segmentation in color images of the retina[J]. IEEE Transactions on Medical Imaging, 2004, 23(4): 501-509. |
[17] | LIN J W, LIAO X W, YU L, et al. Res-UNet based optic disk segmentation in retinal image[J]. Journal of Compu-ters, 2020, 31(3): 183-194. |
[18] | WU X, ZHAO L. Study on iris segmentation algorithm based on dense U-Net[J]. IEEE Access, 2019, 7: 123959-123968. |
[19] | KAUPPI T, KALESNYKIENE V, KAMARAINEN J K, et al. DIARETDB 1 diabetic retinopathy database and evaluation protocol[C]// Proceedings of the British Machine Vision Conference 2007, University of Warwick, Sep 10-13, 2007. Durham: BMVA Press, 2007: 1-10. |
[1] | 吕晓琦, 纪科, 陈贞翔, 孙润元, 马坤, 邬俊, 李浥东. 结合注意力与循环神经网络的专家推荐算法[J]. 计算机科学与探索, 2022, 16(9): 2068-2077. |
[2] | 李珍琦, 王晶, 贾子钰, 林友芳. 融合注意力的多维特征图卷积运动想象分类[J]. 计算机科学与探索, 2022, 16(9): 2050-2060. |
[3] | 夏鸿斌, 肖奕飞, 刘渊. 融合自注意力机制的长文本生成对抗网络模型[J]. 计算机科学与探索, 2022, 16(7): 1603-1610. |
[4] | 李运寰, 闻继伟, 彭力. 高帧率的轻量级孪生网络目标跟踪[J]. 计算机科学与探索, 2022, 16(6): 1405-1416. |
[5] | 张雁操, 赵宇海, 史岚. 融合图注意力的多特征链接预测算法[J]. 计算机科学与探索, 2022, 16(5): 1096-1106. |
[6] | 赵鹏飞, 谢林柏, 彭力. 融合注意力机制的深层次小目标检测算法[J]. 计算机科学与探索, 2022, 16(4): 927-937. |
[7] | 陆仲达, 张春达, 张佳奇, 王子菲, 许军华. 双分支网络的苹果叶部病害识别[J]. 计算机科学与探索, 2022, 16(4): 917-926. |
[8] | 黄思远, 赵宇海, 梁燚铭. 融合图嵌入和注意力机制的代码搜索[J]. 计算机科学与探索, 2022, 16(4): 844-854. |
[9] | 王燕妮, 余丽仙. 注意力与多尺度有效融合的SSD目标检测算法[J]. 计算机科学与探索, 2022, 16(2): 438-447. |
[10] | 蒋光峰, 胡鹏程, 叶桦, 仰燕兰. 基于重构误差的同构图分类模型[J]. 计算机科学与探索, 2022, 16(1): 185-193. |
[11] | 蔡明昕, 孙晶, 王斌. 多角度语义轨迹相似度计算模型[J]. 计算机科学与探索, 2021, 15(9): 1632-1640. |
[12] | 陈俊芬, 张明, 赵佳成, 谢博鋆, 李艳. 结合降噪和自注意力的深度聚类算法[J]. 计算机科学与探索, 2021, 15(9): 1717-1727. |
[13] | 张晗, 贾甜远, 骆方, 张生, 邬霞. 面向网络文本的BERT心理特质预测研究[J]. 计算机科学与探索, 2021, 15(8): 1459-1468. |
[14] | 陈剑南, 杜军平, 薛哲, 寇菲菲. 基于多重注意力的金融事件大数据精准画像[J]. 计算机科学与探索, 2021, 15(7): 1237-1244. |
[15] | 方钧婷, 谭晓阳. 注意力级联网络的金属表面缺陷检测算法[J]. 计算机科学与探索, 2021, 15(7): 1245-1254. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||