Dil-UNet++：基于UNet++的多尺度融合视网膜血管分割网络模型
Dil-UNet++: A Multi-Scale Fusion Retinal Vessel Segmentation Network Model Based on UNet++
DOI: 10.12677/CSA.2024.141007, PDF,  被引量    科研立项经费支持
作者: 米文辉, 佘海州：沈阳航空航天大学，电子工程信息学院，辽宁 沈阳；李 鹤*：沈阳工学院，信息与控制学院，辽宁 抚顺
关键词: 图像分割；视网膜血管；UNet++；空洞卷积；注意力机制；Image Segmentation； Retinal Vessel； UNet++； Dilated Convolution； Attention Module
摘要: 视网膜血管的精确分割是眼部疾病临床诊断的关键步骤，针对眼底视网膜血管结构复杂、形态多样、边缘末端细微导致分割难度高的问题，提出了一种基于UNet++新的眼底视网膜血管分割网络模型(Dil-UNet++)，该方法在传统的UNet++网络基础上使用多层空洞卷积模块来实现特征提取，使网络获得大小可变的特征提取感受野，获得了更好的特征提取性能；同时在网络跳跃连接部分增加注意力机制模块提高了分割目标相关的通道权重和空间权重；并使用卷积模块代替最大池化模块实现下采样，来避免特征传递时血管细节特征丢失。在分割DRIVE视网膜血管图像数据集上的训练与验证结果表明，该模型在Dice系数、准确度和精确度分别达到87.65%、96.99%和92.82%，证明了该网络模型的有效性，从而为眼底视网膜血管图像分割提供了新的方法。
Abstract: The accurate segmentation of retinal blood vessels is a critical step in the clinical diagnosis of ocular disorders. However, the vessel segmentation remains highly challenging due to complex structures, as well as blurred edges and variable morphology. Based on UNet++, we propose a new network model for segmenting retinal vessels (Dil-UNet++). The Dil-UNet++ employs multi-layered void convolutional modules for feature extraction, enabling the network to achieve a flexible feature ex-traction receptive field, thereby enhancing the performance of feature extraction. The addition of the attention mechanism module in the jump connection part of the network improves the channel weight and spatial weight associated with the segmentation target. Instead of maximum pooling, convolution is utilized for down-sampling to avoid loss of vessel details. Based on the segmented DRIVE retinal vessel image dataset, the training and validation results show that the Dice coefficient, ac-curacy, and precision are respectively 87.65%, 96.99%, and 92.82%, which prove the validity of Dil-UNet++ model.
文章引用：米文辉, 李鹤, 佘海州. Dil-UNet++：基于UNet++的多尺度融合视网膜血管分割网络模型[J]. 计算机科学与应用, 2024, 14(1): 54-67. https://doi.org/10.12677/CSA.2024.141007

参考文献

[1] Du, L.L., Liu, H.R., Zhang, L., et al. (2023) Deep Ensemble Learning for Accurate Retinal Vessel Segmentation. Com-puters in Biology and Medicine, 158, Article ID: 106829. [Google Scholar] [CrossRef] [PubMed]
[2] Yi, H. and Tao, D. (2023) Multi-Level Spatial-Temporal and Attentional Information Deep Fusion Network for Retinal Vessel Segmentation. Physics in Medicine and Biology, 68, Article ID: 195026.
[3] Mistry, J. and Ramakrishnan, R.. (2023) The Automated Eye Cancer Detection through Ma-chine Learning and Image Analysis in Healthcare. Journal of Xidian University, 17, 763-772.
[4] Owen, B.J. and Sathyaprakash, B.S. (1998) Matched Filtering of Gravitational Waves from Inspiraling Compact Binaries: Computational Cost and Template Placement. Physical Review D, 60, 727-747. [Google Scholar] [CrossRef] 
[5] Lal, A.M. and Dinesh, A.D. (2014) Abnormality Extraction of MRI Brain Images Using Region Growing Segmentation Techniques.
https://api.semanticscholar.org/CorpusID:212583850
[6] Zhang, L., Jiao, K. and Zhang, L. (2022) Prediction of Blast Furnace Fuel Ratio Based on Back-Propagation Neural Network and K-Nearest Neighbor Algorithm. Steel Re-search International, 93, Article ID: 2200215. [Google Scholar] [CrossRef] 
[7] Jindal, A., Dhir, R. and Rani, R. (2022) Diagonal Features and SVM Classifier for Handwritten Gurumukhi Character Recognition.
https://api.semanticscholar.org/CorpusID:124805943
[8] Shelhamer, E., Long, J. and Darrell, T. (2017) Fully Convolutional Networks for Semantic Segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39, 640-651. [Google Scholar] [CrossRef] 
[9] Ronneberger, O., Fischer, P. and Brox, T. (2015) U-Net: Convolutional Networks for Biomedical Image Segmentation. In: Navab, N., Hornegger, J., Wells, W. and Frangi, A., Eds., Medical Image Computing and Computer-Assisted Intervention—MICCAI 2015, Springer, Cham, 234-241. [Google Scholar] [CrossRef] 
[10] Oktay, O., Schlemper, J. and Folgoc, L.L. (2018) Attention U-Net: Learning Where to Look for the Pancreas.
https://arxiv.org/abs/1804.03999
[11] Zhou, Z., Siddiquee, M.M.R., Tajbakhsh, N. and Liang, J. (2020) UNet++: Redesigning Skip Connections to Exploit Multiscale Features in Image Segmentation. IEEE Transactions on Medical Imaging, 39, 1856-1867. [Google Scholar] [CrossRef] 
[12] Xu, J.C., Wang, S.M., Zhou, Z.J., et al. (2022) Automatic CT image segmentation of maxillary sinus based on VGG network and improved V-Net. International Journal of Computer Assisted Radiology and Surgery, 15, 1457-1465. [Google Scholar] [CrossRef] [PubMed]
[13] Zhao, T., Pan, S., Gao, W., et al. (2021) Attention Unet++ for Lightweight Depth Estimation from Sparse Depth Samples and a Single RGB Image. The Visual Computer, 38, 1619-1630. [Google Scholar] [CrossRef] 
[14] 浦秀丽, 刘翔, 汤显, 等. 基于边缘监督的肝部超声图像包膜分割网络[J]. 中国医学物理学杂志, 2022, 39(10): 1255-1262.
[15] Siddique, N., Paheding, S. and Devabhaktuni, V. (2021) U-Net and Its Variants for Medical Image Segmentation: A Review of Theory and Applications. IEEE Access, 9, 82031-82057. [Google Scholar] [CrossRef] 
[16] Li, C., et al. (2020) Attention Unet++: A Nested Attention-Aware U-Net for Liver CT Image Segmentation. 2020 IEEE International Conference on Image Processing (ICIP), Abu Dhabi, 25-28 October 2020, 345-349. [Google Scholar] [CrossRef] 
[17] Wang, Z. and Ji, S. (2021) Smoothed Dilated Convolutions for Improved Dense Prediction. Data Mining and Knowledge Discovery, 35, 1470-1496. [Google Scholar] [CrossRef] 
[18] Xu, C., Wang, X. and Zhang, S. (2022) Dilated Convolution Capsule Network for Apple Leaf Disease Identification. Frontiers in Plant Science, 13, Article 1002312. [Google Scholar] [CrossRef] [PubMed]
[19] 黄晓鸣, 何富运, 唐晓虎, 等. U-Net及其变体在医学图像分割中的应用研究综述[J]. 中国生物医学工程学报, 2022, 41(5): 567-576.
[20] 殷宁波, 黄冕, 刘利军, 等. MS-UNet++: 基于改进UNet++的视网膜血管分割[J]. 光电子∙激光, 2021, 32(1): 35-41. [Google Scholar] [CrossRef] 
[21] Pesch, I.S., Bestelink, E., de Sagazan, O., Mehonic, A. and Sporea, R.A. (2022) Multimodal Transistors as ReLU Activation Functions in Physical Neural Network Classifiers. Sci-entific Reports, 12, Article No. 670. [Google Scholar] [CrossRef] [PubMed]
[22] Wang, Z.Y. (2021) Design of Fundus Vascular Segmentation System Based on Improved UNet++ Algorithm. Master’s Thesis, Yanbian University, Yanji. [Google Scholar] [CrossRef] 
[23] Ibtehaz, N. and Rahman, M.S. (2020) MultiResUNet: Rethinking the U-Net Architecture for Multimodal Biomedical Image Segmentation. Neural Networks, 121, 74-87. [Google Scholar] [CrossRef] [PubMed]
[24] 王忠源. 基于改进型UNet++算法的眼底血管分割系统设计[D]: [硕士学位论文]. 延边: 延边大学, 2021.[CrossRef] 
[25] Ali, Z., Irtaza, A. and Maqsood, M. (2021) An Efficient U-Net Framework for Lung Nodule Detection Using Densely Connected Dilated Convolutions. The Journal of Super-computing, 78, 1602-1623. [Google Scholar] [CrossRef] 
[26] Sun, X.F., Li, J.M., Ma, J.L., et al. (2021) Segmentation of Overlapping Chromosome Images Using U-Net with Improved Dilated Convolutions. Journal of Intelligent & Fuzzy Systems, 40, 5653-5668. [Google Scholar] [CrossRef] 
[27] Chen, Q., Xie, W., Zhou, P., et al. (2021) Multi-Crop Convolutional Neural Networks for Fast Lung Nodule Segmentation. IEEE Transactions on Emerging Topics in Computational Intel-ligence, 6, 1190-1200. [Google Scholar] [CrossRef] 
[28] Vigneron, V., Maaref, H. and Syed, T.Q. (2021) A New Pool-ing Approach Based on Zeckendorf’s Theorem for Texture Transfer Information. Entropy, 23, Article 279. [Google Scholar] [CrossRef] [PubMed]
[29] Natalia, N., Isidoros, P. and Ioannis, D. (2023) A Convolutional Autoen-coder Approach for Boosting the Specificity of Retinal Blood Vessels Segmentation. Applied Sciences, 13, Article 3255. [Google Scholar] [CrossRef] 
[30] 徐微, 汤俊伟, 张驰. DC-CBAM-UNet++网络的肺结节图像分割方法[J]. 软件导刊, 2023, 22(7): 125-130.
[31] Du, H.W., Zhang, X.Y., Song, G., et al. (2022) Retinal Blood Vessel Segmentation by Using the MS-LSDNet Network and Geometric Skeleton Reconnection Method. Computers in Biology and Medicine, 153, Article ID: 106416. [Google Scholar] [CrossRef] [PubMed]
[32] Li, B., Wu, F., et al. (2022) CA-Unet++: An Improved Structure for Medical CT Scanning Based on the Unet++ Architecture. International Journal of Intelligent Systems, 37, 8814-8832. [Google Scholar] [CrossRef] 
[33] Zhang, T., Li, J., Zhao, Y., et al. (2022) MC-UNet Mul-ti-Module Concatenation Based on U-Shape Network for Retinal Blood Vessels Segmentation. arXiv: 2204.03213.
[34] Marcellino, Cenggoro, T.W. and Pardamean, B. (2022) UNET++ with Scale Pyramid for Crowd Counting. ICIC Express Letters, 16, 75-82.
[35] Huang, A., Jiang, L., Zhang, J. and Wang Q. (2022) Atten-tion-VGG16-UNet: A Novel Deep Learning Approach for Automatic Segmentation of the Median Nerve in Ultrasound Images. Quantitative Imaging in Medicine and Surgery, 12, 3138-3150. [Google Scholar] [CrossRef] [PubMed]
[36] Feng, J., Deng, J., Li, Z., Sun, Z., et al. (2020) End-to-End Res-Unet Based Reconstruction Algorithm for Photoacoustic Imaging. Biomedical Optics Express, 11, 5321-5340. [Google Scholar] [CrossRef]