位置感知循环卷积与多尺度输入的视网膜血管分割方法

doi:10.3778/j.issn.1002-8331.2305-0227

摘要/Abstract

摘要： 针对视网膜图像血管纹理复杂，微小血管极多，成像对比度低的问题，提出一种结合位置感知循环卷积（position aware circular convolution，ParC）、多尺度分辨率输入的视网膜血管分割方法。使用带有普通卷积、位置感知循环卷积、ECA（efficient channel attention）注意力的卷积模块（ParC-ECA block）来充分提取输入眼底图像的全局、局部特征信息；在级联的下采样路径中，提出多尺度输入模块（multi-scale input block）来对每一层级的特征信息进行加强，找回丢失的细节信息，避免因细节丢失而引起的网络性能下降；在跳跃连接中使用残差双注意力模块（residual spatial channel attention block，RSCA），在保持网络每一层级原始特征传递的基础上，对其进行背景干扰噪声过滤和血管特征强化，进一步提升分割性能。提出的方法在DRIVE数据集和CHASE_DB1数据集上进行了实验，其AUC分别为98.53%和98.81%，ACC分别为95.81%和96.84%，F1-score分别为83.55%和83.39%。实验结果表明所提方法优于现有主流分割方法，特别是在微小血管的分割表现方面较为突出。

关键词: 视网膜血管分割, 多尺度输入, 位置感知循环卷积, 注意力机制

Abstract: A retinal vessel segmentation method combining position-aware circular convolution and multi-scale resolution input is proposed for the problem of complex retinal image vessel texture, extremely many tiny vessels, and low imaging contrast. The convolution module（ParC-ECA Block） with ordinary convolution, position-sensitive circular convolution, and ECA（efficient channel attention） attention is used to fully extract the global and local feature information of the input fundus image. In the cascaded downsampling path, the multi-scale input module（multi-scale input block） is proposed to enhance the feature information at each level, retrieve the lost detail information, and avoid the network performance degradation caused by the detail loss. In the jump connection using the residual dual attention module（residual spatial channel attention block, RSCA） to further improve the segmentation performance by performing background interference noise filtering and vascular feature enhancement on the basis of maintaining the original feature transfer at each level of the network. The proposed method is experimented on the DRIVE dataset and CHASE_DB1 dataset, with AUCs of 98.53% and 98.81%, ACCs of 95.81% and 96.84%, and F1-scores of 83.55% and 83.39%, respectively. The experimental results show that the proposed method outperforms the existing mainstream segmentation methods, especially in the segmentation of tiny vessels.

Key words: retinal vessel segmentation, multi-scale input, position-aware circular convolution, attention mechanism

江中川, 吴云. 位置感知循环卷积与多尺度输入的视网膜血管分割方法[J]. 计算机工程与应用, 2023, 59(21): 242-250.

JIANG Zhongchuan, WU Yun. Retinal Vessel Segmentation Method Based on Position-Aware Circular Convolution with Multi-Scale Input[J]. Computer Engineering and Applications, 2023, 59(21): 242-250.

参考文献

[1] RASMUSSEN M L，BROE R，FRYDKJAER-OLSEN U，et al.Retinal vascular geometry and its association to microvascular complications in patients with type 1 diabetes：the Danish Cohort of Pediatric Diabetes 1987（DCPD1987）[J].Graefe’s Archive for Clinical and Experimental Ophthalmology，2017，255：293-299.
[2] TRIANTAFYLLOU A，AL-DIRI B，ANYFANTI P，et al. 7D.08：detecting hypertensive retinopathy using retinal vascular geometry[J].Journal of Hypertension，2015，33：e102.
[3] DASH J，BHOI N.A thresholding based technique to extract retinal blood vessels from fundus images[J].Future Computing and Informatics Journal，2017，2（2）：103-109.
[4] SREEJINI K S，GOVINDAN V K.Improved multiscale matched filter for retina vessel segmentation using PSO algorithm[J].Egyptian Informatics Journal，2015，16（3）：253-260.
[5] HASSAN G，EL-BENDARY N，HASSANIEN A E，et al.Retinal blood vessel segmentation approach based on mathematical morphology[J].Procedia Computer Science，2015，65：612-622.
[6] CHENG Y，HU X，WANG J，et al.Accurate vessel segmentation with constrained B-snake[J].IEEE Transactions on Image Processing，2015，24（8）：2440-2455.
[7] SULE O O.A survey of deep learning for retinal blood vessel segmentation methods：taxonomy，trends，challenges and future directions[J].IEEE Access，2022，10：38202-38236.
[8] LONG J，SHELHAMER E，DARRELL T.Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2015：3431-3440.
[9] RONNEBERGER O，FISCHER P，BROX T.U-net：convolutional networks for biomedical image segmentation[C]//18th International Conference on Medical Image Computing and Computer-Assisted Intervention-（MICCAI 2015），Munich，Germany，October 5-9，2015：234-241.
[10] ZHOU Z，RAHMAN SIDDIQUEE M M，TAJBAKHSH N，et al.Unet++：a nested u-net architecture for medical image segmentation[C]//4th International Workshop on Deep Learning in Medical Image Analysis and 8th International Workshop on Multimodal Learning for Clinical Decision Support，September 20，2018：3-11.
[11] YUE K，ZOU B，CHEN Z，et al.Retinal vessel segmentation using dense U-net with multiscale inputs[J].Journal of Medical Imaging，2019，6（3）：034004.
[12] OKTAY O，SCHLEMPER J，FOLGOC L L，et al.Attention u-net：learning where to look for the pancreas[J].arXiv：1804.03999，2018.
[13] ZHANG H，HU W，WANG X.Parc-net：position aware circular convolution with merits from convnets and transformer[C]//17th European Conference on Computer Vision，Tel Aviv，Israel，October 23-27，2022.Cham：Springer Nature Switzerland，2022：613-630.
[14] WANG Q，WU B，ZHU P，et al.ECA-net：efficient channel attention for deep convolutional neural networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition，2020：11534-11542.
[15] 周飞燕，金林鹏，董军.卷积神经网络研究综述[J].计算机学报，2017，40（6）：1229-1251.
ZHOU F Y，JIN L P，DONG J.Review of convoluntional neural network[J].Chinese Journal of Computers，2017，40（6）：1229-1251.
[16] YU F，KOLTUN V.Multi-scale context aggregation by dilated convolutions[J].arXiv：1511.07122，2015.
[17] DAI J，QI H，XIONG Y，et al.Deformable convolutional networks[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision，Venice，Oct 22-29，2017.Washington：IEEE Computer Society，2017：764-773.
[18] 刘卫光，刘东，王璐.可变形卷积网络研究综述[J].计算机科学与探索，2023，17（7）：1549-1564.
LIU W G，LIU D，WANG L.Survey of deformable convoluntional networks[J].Journal of Frontiers of Computer Science and Technology，2023，17（7）：1549-1564.
[19] DENG L，GONG Y，LU X，et al.Focus-enhanced scene text recognition with deformable convolutions[C]//Proceedings of the 2019 IEEE 5th International Conference on Computer and Communications，Chengdu，Dec 6-9，2019.Washington：IEEE Computer Society，2019：1685-1689.
[20] ZHUANG J.LadderNet：multi-path networks based on U-Net for medical image segmentation[J].arXiv：1810. 07810，2018.
[21] ZHOU Y，CHEN Z，SHEN H，et al.A refined equilibrium generative adversarial network for retinal vessel segmentation[J].Neurocomputing，2021，437：118-130.
[22] GUO S.CSGNet：cascade semantic guided net for retinal vessel segmentation[J].Biomedical Signal Processing and Control，2022，78：103930.
[23] HUSSAIN S，GUO F，LI W，et al.DilUnet：a U-net based architecture for blood vessels segmentation[J].Computer Methods and Programs in Biomedicine，2022，218：106732.
[24] HU J，SHEN L，SUN G.Squeeze-and-excitation networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2018：7132-7141.
[25] WOO S，PARK J，LEE J Y，et al.Cbam：convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision（ECCV），2018：3-19.
[26] OWEN C G，RUDNICKA A R，MULLEN R，et al.Measuring retinal vessel tortuosity in 10-year-old children：validation of the computer-assisted image analysis of the retina（CAIAR） program[J].Investigative Ophthalmology & Visual Science，2009，50（5）：2004-2010.
[27] 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.
[28] 李兰兰，张孝辉，牛得草，等.深度学习在视网膜血管分割上的研究进展[J].计算机科学与探索，2021，15（11）：2063-2076.
LI L L，ZHANG X H，NIU D C，et al.Research progress of deep learning on retinal vessel segmentation[J].Journal of Frontiers of Computer Science and Technology，2021，15（11）：2063-2076.
[29] LI L，VERMA M，NAKASHIMA Y，et al.Iternet：retinal image segmentation utilizing structural redundancy in vessel networks[C]//Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision，2020：3656-3665.
[30] YANG D，LIU G，REN M，et al.A multi-scale feature fusion method based on u-net for retinal vessel segmentation[J].Entropy，2020，22（8）：811.
[31] JIN Q，MENG Z，PHAM T D，et al.DUNet：a deformable network for retinal vessel segmentation[J].Knowledge-Based Systems，2019，178：149-162.
[32] ZHANG T，LI J，ZHAO Y，et al.Mc-unet multi-module concatenation based on u-shape network for retinal blood vessels segmentation[J].arXiv：2204.03213，2022.