眼睛是不可或缺的人体器官，其健康与人类生活密不可分。医生在进行眼部疾病以及某些人体器官状态诊断时，主要依靠对眼底视网膜图像的判读。眼底视网膜图像的血管分割是医生专家诊治患者状态的决定性依赖条件，对于临床医学非常有意义，因此研究如何有效的进行眼底视网膜血管的分割是必要的。近年来，随着深度学习在计算机视觉领域的迅猛发展，已有许多基于深度学习的方法应用于眼底视网膜血管分割中，但在实际应用中由于视网膜血管病变和拍摄光照等因素可能造成血管与背景对比度低下，从而导致细微血管不能很好识别，繁杂曲度形态下的血管欠分割、过分割以及分割灵敏度指标较低等问题。本文基于深度学习提出了两种不同的眼底视网膜图像血管分割算法，论文的主要研究内容如下：

（1）设计研究了基于编解码结构的多特征融合视网膜血管分割模型。针对眼底视网膜图像血管分割中血管细节部分丢失或断裂导致准确率不高的问题，在编解码结构的U型网络基础上提出了一种新的视网膜血管分割算法。通过引入残差模块、短跳跃连接使得高低特征相融合，更好的完成信息传递；引入空洞卷积和注意力机制，增大感受野，将特征信息强化，提高网络泛化能力。将这些改进模块巧妙结合，提高网络模型的整体性能。

（2）设计研究了基于生成式对抗网络的改进视网膜血管分割模型。针对眼底视网膜血管分割在血管分布复杂处细微血管的欠分割、过分割现象以及灵敏度低下的问题，提出了一种基于生成式对抗网络的改进血管分割算法模型。该模型中的生成器部分在前一章提出的改进U型编解码结构的基础上，联合Ladder Net的多对编解码组合结构，使两个原本的改进U型编解码网络并行相互连接，将模型拓展为W型网络结构，获得了更多的信息流路径，从而在特征图上得到了更多的补充，使得网络模型更好的利用上下文信息，进而提高分割效率与精度；判别器部分在原始卷积网络中加入残差块结构，以解决网络退化的影响。通过两个改进网络模型交替训练，模型整体对视网膜血管分割精准度大幅度提升。

本文在眼底图像DRIVE、CHASE-DAB1、STARE三个数据集进行两种模型的训练和测试。实验结果表明，两种模型在三个数据集上灵敏度分别达到0.8093和0.8326、0.8112和0.8238、0.8118和0.8394，研究提出的两种改进网络模型在细微血管的分割上具有精准的表现，并且对视网膜血管分割像素的灵敏度有非常明显的提升，网络模型整体效果较优。

The eye is a vital part of the human body. Its health is inextricably linked to human life. Doctors rely heavily on the interpretation of fundus retinal pictures when diagnosing eye disorders and various organ conditions. The essential condition for doctors to diagnose patients diseases is fundus vascular segmentation, which is of great significance for clinical medicine. As a result, it is necessary to study segment fundus retinal vessels for doctors to diagnose diseases effectively. Many methods based on deep learning have been applied to fundus retinal vascular segmentation in recent years, thanks to the rapid growth of deep learning in the field of computer vision. However, the contrast between blood vessels and background may be low in practice due to retinal vascular disease, shooting light, and other factors, resulting in poor recognition of fine blood vessels, under segmentation of blood vessels in complex curvature forms over segmentation, and low segmentation sensitivity. This research provides two distinct vascular segmentation techniques for fundus retinal images based on deep learning. The main research contents of this paper are as follows:

A multi feature fusion retinal vessel segmentation model based on codes structure is designed. Aiming at the problem of low accuracy caused by the loss or fracture of vascular details in vascular segmentation of fundus retinal image. A new retinal vessel segmentation algorithm is proposed based on the U-shaped network of encoding and decoding structure. The model completes information transfer better by introducing residual module and short hop connection to fuse high and low features. In addition, the model introduces void convolution and attention mechanism to expand the receptive field, enhance feature information and improve the generalization ability of the network. The overall performance of the network model has been improved by combining these improved modules cleverly.

(2) A retinal vascular segmentation model based on conditional generation countermeasure network is designed. Aiming at the problems of under segmentation, over segmentation and low sensitivity of micro vessels in complex distribution in fundus retinal vascular segmentation. An improved blood vessel segmentation algorithm model based on generative countermeasure network is proposed. The generator part of the model is based on the encoding and decoding structure and the improved U-shaped structure. On this basis, combined with the multi pair encoding and decoding structure in ladder net, the two original improved U-shaped network structures are connected in parallel and expanded into W-shaped network structure, which has been more information flow paths. As a result, the model gets more supplements on the characteristic graph to make use of context information better. The accuracy and efficiency of retinal segmentation are improved. In the discriminator part, the original convolution network added the residual block structure to solve the influence of network degradation. The accuracy of retinal blood vessel segmentation of the model as a whole is improved greatly ,through the alternate training with two improved network models.

In this paper, the two models are trained and tested on the fundus image DRIVE, STARE and CHASE-DB1 data sets. The results of the experiments reveal that the sensitivities of the two models on the three data sets are 0.8093 and 0.8326, 0.8112 and 0.8238, 0.8118 and 0.8394 respectively. The suggested two upgraded network models are accurate in segmenting micro vasculature. The sensitivity of retinal blood vessel segmentation pixels has improved greatly, and the network model's overall effect has improved.

[1]梁礼明, 盛校棋, 蓝智敏, 等. 自适应尺度信息的U型视网膜血管分割算法[J]. 光学学报, 2019, 39(8): 499-512.

[2]Zhang W , Wang X , You W , et al. RESLS: Region and Edge Synergistic Level Set Framework for Image Segmentation[J]. IEEE Transactions on Image Processing, 2019, 29:57-71.

[3]Zhao R, Li Q, Wu J, et al. A nested U-shape network with multi-scale up-sample attention for robust retinal vascular segmentation[J]. Pattern Recognition, 2021, 120(4): 68-77.

[4]Boudegga H, Elloumi Y, Akil M, et al. Fast and efficient retinal blood vessel segmentation method based on deep learning network[J]. Computerized Medical Imaging and Graphics, 2021, 90(4): 169-208.

[5]祁炜. 基于导向复小波的视网膜图像增强与血管监督分割技术研究[D]. 武汉:华中科技大学, 2019.

[6]Arias M, Santos M, Borrero P, et al. A new deep learning method for blood vessel segmentation in retinal images based on convolutional kernels and modified U-Net model[J]. Computer Methods and Programs in Biomedicine, 2021, 23(4): 167-178.

[7]Orando J, Prokofyeva E,Blaschko M. A discriminativly trained fully connected conditional random field model for blood vessel segmentation in fundus images[J]. IEEE Transactions on Biomedical Engineering, 2017, 64(1): 16-27.

[8]Liskowski P, Krawiec K. Segmenting Retinal Blood Vessels With Deep Neural Networks[J]. IEEE Transactions on Medical Imaging, 2017, 35(11): 2369-2380.

[9]Dasgupta A, Singh S. A fully convolutional neural network based structured prediction approach towards the retinal vessel segmentation[J]. Medical Image Analysis, 2017, 45(33): 248-251.

[10]Lahiri A, Ayush K, Biswas P, et al. Generative adversarial learning for reducing manual annotation in semantic segmentation on large scale microscopy images automated vessel segmentation in retinal fundus image as test case[C]//IEEE Conference on Computer Vision and Pattern Recognition Workshops(CVPRW).Singapore, 2017: 42-48.

[11]Ngo L, Han J. Multi-level deep neural network for efficient segmentation of blood vessels fundus images [J]. Electronics Letters, 2018, 53(16): 2274-2282.

[12]Adeyinka A, Adebiyi M, Akande N, et al. A Deep Convolutional Encoder-Decoder Architecture for Retinal Blood Vessels Segmentation[C]//Computational Science and Its Applications ICCSA. Russia:Springer, 2019:180-189.

[13]Morano J, Hervella L, Barreira N, et al. Multi modal Transfer Learning-based Approaches for Retinal Vascular Segmentation[J]. Electrical Engineering and Systems Science, 2020, 23(7): 632-645.

[14]Chen Q, Tan F, Wu Y, et al. Characteristics of Retinal Structural and Microvascular Alterations in Early Type 2 Diabetic Patients[J]. Investigative Ophthalmology & Visual Science, 2018, 59(5): 210-217.

[15]Hu K, Zhang Z, Niu X, et al. Retinal vessel segmentation of color fundus images using multi scale convolutional neural network with an improved cross-entropy loss function[J]. Neurol Computing, 2018, 309(10): 179-191.

[16]Guo Y. A Retinal Vessel Detection Approach Using Convolution Neural Network with Reinforcement Sample Learning Strategy[J]. Measurement, 2018, 125(5): 586-591.

[17]Xiao X, Lian S, Luo Z, et al. Weighted Res-U Net for High-Quality Retina Vessel Segmentation[C]//Proceedings of the 2018 9th International Conference on Information Technology in Medicine and Education (ITME). Computer Society, 2018: 327-331.

[18]Wang Y, Yin W, Zheng y. Hierarchical retinal blood vessel segmentation based on feature and ensemble learning[J]. Electronics Letters, 2019, 149(2): 708-717.

[19]Jiang Y, Tan N, Peng T, et al. Retinal Vessels Segmentation Based on Dilated Multi-

Scale Convolutional Neural Network[J]. IEEE Access, 2019, 38(6): 231-252.

[20]汪维华, 张景中, 吴文渊. 改进的形态学与OTSU相结合的视网膜血管分割[J]. 计算机应用研究, 2019, 23(7): 2228-2231.

[21]Srinidhi C,Aparna P,Rajan J.Recent Advancements in Retinal Vessel Segmentation[J]. Journal of Medical Systems, 2018, 41(4): 70-73.

[22]Yang J, Huang M, Fu J, et al. Frangi based multi-scale level sets for retinal vascular segmentation[J]. Computer Methods Programs in Biomedicine, 2020, 197(3): 2011-2023.

[23]Odstreilik J, Kolar R, Budai A, et al. Retinal vessel segmentation by improved matched filtering: evaluation on a new high-resolution fundus image database[J]. IET Image Processing, 2013, 7(4): 373-383.

[24]Na T, Xie J, Zhao Y, et al. Retinal Vascular Segmentation Using Super pixel‐based Line Operator and Its Application to Vascular Topology Estimation[J]. Medical Physics, 2018, 34(3):421-433..

[25]Chang C, Chu A, Meyer M, et al. Three-dimensional imaging coupled with topological quantification uncovers retinal vascular plexuses undergoing obliteration[J]. Theranostics, 2021, 11(3): 1162-1175.

[26]Kumar M,Remagino P,Hoppe A,et al.Blood vessel segmentation on methodologies in retinal image survey[J].Computer Methods and Programs in Medicine, 2020, 108(1): 407-433.

[27]朱辉, 秦品乐. 基于多尺度特征结构的U-Net肺结节检测算法[J]. 计算机工程, 2019, 45(4): 254-261.

[28]Cheng L, Lyu X, Zhang S. The RETA Benchmark for Retinal Vascular Tree Analysis[J]. Ar-Xiv prints, 2021, 28(9): 1488-1497.

[29]Ghosh K, Ghosh A. A novel retinal image segmentation using SVM boosted convolutional neural network for detection[J]. Biomedical Signal Processing and Control, 2021, 68(2): 2481-2495.

[30]Feng S, Zhuo Z, Pan D, et al. Cc-Net A cross-connected convolutional network for segmenting retinal vessels using multi-scale features[J]. Neurol Computing, 2020, 392(5): 268-276.

[31]Azar A, Malek J. 3D surface reconstruction of retinal vascular structures[J]. International Journal of Modelling Identification and Control, 2016, 26(4): 303-316.

[32]Rahman A, Biswal B, Geetha P, et al. Robust segmentation of vascular network using deeply cascaded AReU-Net[J]. Biomedical Signal Processing and Control, 2021, 69(3): 105-108.

[33]郭臻, 陈远知. 图像阈值分割算法研究[J]. 中国传媒大学学报, 2018, 12(152): 77-82.

[34]Trinh M , Kalloniatis M , Nivison L. Radial papillary capillary plexus sparing and underlying retinal vascular impairment in intermediate Age-Related macular Degeneration[J]. Investigative Ophthalmology&Visual Science, 2021, 62(4): 6345-6360.

[35]Wang H, Hu H, Gregori G, et al. The effect of software versions on the measurement of retinal vascular densities using optical coherence tomography angiography[J]. Current Eye Research, 2020, 21(12): 63-66.

[36]Zhao R, Li Q, Wu J, et al. A nested U-shape network with multi-scale up-sample attention for robust retinal vascular segmentation[J]. Pattern Recognition, 2021, 120(4): 107-118.

[37]孟琳, 刘静, 曹慧, 等. 基于Frangi滤波器和OTSU视网膜血管分割[J]. 激光与光电子学进展, 2019, 653(18): 127-133.

[38]Zhao S, Chen Z. Image segmentation method integrating global binarization and edge detection[J]. Journal of Computer Aided Design and Graphics, 2018, 13(2): 118-121.

[39]Pizer S, Ambun D,Austin, et al. Adaptive histogram equalization and its variation Computer[J]. Computer Vision Graphics and Image Processing, 2018, 39(3): 355-368.

[40]Zhu C, Zou B, Zhao R, et al. Retinal vessel segmentation in colour fundus images using extreme learning machine[J]. Computerized Medical Imaging and Graphics, 2017, 32(2): 30-32.

[41]Rouco J. Deep Multi-segmentation approach for the joint classification and segmentation of the retinal arterial and venous trees in color fundus images[J]. Engineering Proceedings, 2021, 7(13): 168-176.

[42]He R,Jiang G.Retinal Vessel Segmentation Multi-scale Deep Guidance[J]. Pacific-Rim Conference on Multimedia, 2019, 31(6): 212-224.

[43]Musial G, Queener H, Adhikari S, et al. Automatic segmentation of retinal capillaries in adaptive optics scanning laser ophthalmologist perfusion images using a convolutional neural network[J]. Transitional Vision Science & Technology, 2020, 9(2):43-48.

[44]Niemeijer M, Gimneken M, et al. Retinopathy online challenge automatic detection of micro-aneurysms in digital color fundus photo graphs[J].IEEE Transactions on Medical Imaging, 2018, 29(1): 185-195.

[45]Prokofyeva E, Blaschko B. A discriminativly trained fully connected conditional random field model for blood vessel segmentation in fundus images[J]. IEEE Transcriptions on Biomedical Engineering, 2019, 64(1): 16-27.

[46]Na T, Xie J , Zhao Y, et al. Retinal vascular segmentation using super pixel‐based line operator and application to vascular topology estimation[J]. Medical Physics, 2018, 16(5): 15-26.

[47]Du X, Wang J, Sun W Z. U-Net retinal blood vessel segmentation algorithm based on improved pyramid pooling method and attention mechanism[J]. Physics in Medicine & Biology, 2021, 66(17): 640-651.

[48]Sathananthavathi V,Indumathi G,Swetha R.Parallel Architecture of Fully Convolution Neural Network for Retinal Vessel Segmentation[J].Journal of Digital Imaging, 2020, 33(1): 168-180.

[49]高宏杰. 基于改进U型网络的眼底图像血管分割[J]. 中国生物医学工程学报, 2019, 38(1): 1-8.

[50]Royedhury S, Koozeka D, Parhi K. Blood vessel segmentation of fundus images by major vessel extraction and image classification[J]. IEEE Biomedical Health Inform, 2019, 19(3): 1118-1128.

[51]Chen T, Tang C, Lei K. Multi-scale retinal vessel segmentation based on fully convolutional neural network[J]. Journal of Optics, 2021, 39(2): 21-25.

[52]Long J, Shelhamer E, Trevor D. Fully convolutional networks for semantic segmentation[J].IEEE Conference on Computer Vision and Pattern Recognition, 2018, 25(8): 3431-3440.

[53]Tang M, Xiao L, Zhou J. Automatic retinal vascular segmentation method based on densely connected convolution neural network[J]. Journal of Chengdu University of Information Technology, 2018, 40(4) :834-848.

[54]Yang J, Huang M, Fu J, et al. Frangi based multi-scale level sets for retinal vascular segmentation[J]. Computer Methods and Programs in Biomedicine,2020,197(3):105-152.

[55]Oktay C, Schlemper J, Folgoc L, et al. Attention U-Net:learning look for the pancreas[J]. Medical Image Analysis, 2020, 38(2): 172-183.

[56]GoodFellow I, Mirza M, et al. Generative Adversarial Networks [J]. Proceedings of the Conference on Advances in Neural Information Processing Systems, 2019, 34(23): 2672-2680.

[57]Liang L, Sheng X, Lan Z, et al. U-shaped retinal vessel segmentation algorithm based on adaptive scale information[J]. Computer Engineering, 2019, 39(8): 81-84.

[58]Dharmawan D, Li D, Ngo P, et al. A New Hybrid Algorithm for Retinal Vessels Segmentation on Fundus Images[J]. IEEE Access, 2019, 43(9):121-133.

[59]Zhuang J. Ladder Net: Multi-path networks based on U-Net for medical image segmentation[J]. Proceedings of SPIE for Optical Engineering, 2019, 53(7): 648- 656.

[60]Li Q, Feng B, Zhang H,et al. A Cross Modality Learning Approach for Vessel Segmentation in Retinal Images[J]. IEEE Transactions on Medical Imaging, 2017, 25(1):109-118.

[61]Khan K, Khaliq A, Shahid M. A Novel Fast GLM Approach for Retinal Vascular Segmentation and Denoising[J]. Journal of Information Science and Engineering, 2017, 33(6):1611-1627.

[62]Guo Y, Budak U, Vespa L, et al. A Retinal Vessel Detection Approach Using Convolution Neural Network with Reinforcement Sample Leaning Strategy [J]. Measurement, 2018, 125(65): 586-591.

[63]Oliveira A, Pereira S, Silva C. Retinal vessel segmentation based on fully convolutional neural networks[J]. Expert Systems with Applications, 2018, 11(27): 229-242.

[64]Alom Z, Hasan M, Yakopcic C.Recurrent Residual Convolutional Neural Network based on U-Net (R2U-Net) for Medical Image Segmentation[C]// International Conference on International.Conference on Machine Learning. France: Seattle, 2019: 2212-2310.

[65]Feng Z, Yang J, Yao L. Patch-based fully convolutional neural network with skip connections for retinal blood vessel segmentation [J]. IEEE International Conference on Image Processing, 2019, 36(21): 1742-1746.

[66]Son J, Park S J, Jung K H. Towards accurate segmentation of retinal vessels and the in fundus images with generative adversarial networks[J]. Journal of Digital Imaging, 2018, 32(67): 499-512.

[67]Shi Z, Liu J. Fundus Image Generation based on Generative Adversarial Networks and Deep Learning for Retinal vessel Segmentation[C]//2019 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).USA:Long Beach, 2019:10437-10442.

[68]Wu H, Wang C, Chen C, et al. Association between retinal vascular geometric changes and cognitive impairment: A systematic review and meta-analysis[J]. Journal of Clinical Neurology, 2020, 16(1): 19-27.

[69]Yan Z, Yang X, Cheng K T. Joint segment-level and pixel-wise losses for deep learning based retinal vessel segmentation [J]. IEEE Transactions on Biomedical Engineering, 2019, 65(9): 1912-1923.

[70]Hu K, Zhang Z, Niu X, et al. Retinal vessel segmentation of color fundus images using multi scale convolutional neural network with an improved cross entropy loss function [J]. Neurol Computing, 2018, 309(78): 179-191.