基于多尺度注意力的视网膜血管分割方法研究

doi:10.11896/jsjkx.241200112

摘要/Abstract

摘要： 在医学图像分割中,视网膜血管分割对于眼科疾病的早期诊断与治疗是很重要的。视网膜血管分割不仅有助于诊断糖尿病视网膜病变、青光眼、动脉硬化等疾病,还在分析眼部血管形态、血流动力学等方面具有广泛的应用。但是现有方法在处理视网膜细小血管和血管边缘时还无法精确分割,在类别不平衡、血管形态复杂性和有限训练样本等方面仍然受到限制。为了提高血管分割精度并降低误判率,提出了一种基于多尺度注意力的视网膜血管分割模型(MDAF-Net)。该模型通过引入多尺度动态卷积来自适应地调整对不同尺度血管的关注度,缓解了细小血管提取不足的问题,结合通道和空间注意力机制优化特征融合,增强了模型对细节特征的提取能力,采用多尺度特征融合策略,提升了在血管形态复杂性下的分割效果。MDAF-Net在DRIVE和CHASE_DB1数据集上验证模型效果,得到Dice系数为0.764、MIoU为78.3%(DRIVE)和Dice系数为0.820、MIoU为82.5%(CHASE_DB1)。实验结果表明,MDAF-Net在分割精度和假阳性率控制方面具有显著优势,解决了传统方法在细小血管分割、类别不平衡和假阳性等方面的局限。

关键词: 多尺度, 动态卷积, 注意力融合, 特征提取, 视网膜血管分割

Abstract: In medical image segmentation,retinal vessel segmentation is very important for the early diagnosis and treatment of ophthalmic diseases.Retinal vessel segmentation is not only helpful for the diagnosis of diseases such as diabetic retinopathy,glaucoma,and arteriosclerosis,but also has wide applications in analyzing ocular vascular morphology and hemodynamics.How-ever,existing methods cannot accurately segment small retinal blood vessels and blood vessel edges,and are still limited in terms of class imbalance,complexity of blood vessel morphology,and limited training samples.In order to improve the accuracy of blood vessel segmentation and reduce the false positive rate,this paper proposes a retinal vessel segmentation model based on multi-scale attention(MDAF-Net).The model introduces multi-scale dynamic convolution to adaptively adjust the attention to blood vessels of different scales,alleviates the problem of insufficient extraction of small blood vessels,combines channel and spatial attention mechanisms to optimize feature fusion,enhances the model’s ability to extract detailed features,and adopts a multi-scale feature fusion strategy to improve the segmentation effect under the complexity of blood vessel morphology.MDAF-Net verifies the model effect on the DRIVE and CHASE_DB1 datasets,and obtains a Dice coefficient of 0.764 and an MIoU of 78.3%(DRIVE) and a Dice coefficient of 0.820 and an MIoU of 82.5%(CHASE_DB1).The experimental results show that MDAF-Net has significant advantages in segmentation accuracy and false positive rate control,and solves the limitations of traditional methods in small blood vessel segmentation,category imbalance and false positives.

Key words: Multi-scale, Dynamic convolution, Attention fusion, Feature extraction, Retinal vessel segmentation

中图分类号:

TP391

朱思凡, 朱国胜. 基于多尺度注意力的视网膜血管分割方法研究[J]. 计算机科学, 2025, 52(11A): 241200112-10. https://doi.org/10.11896/jsjkx.241200112

ZHU Sifan, ZHU Guosheng. Retinal Vessel Segmentation Based on Multi-scale Attention[J]. Computer Science, 2025, 52(11A): 241200112-10. https://doi.org/10.11896/jsjkx.241200112

参考文献

[1]ODURO J K,OKYERE J,NYADOR J.Risky health behaviours and chronic conditions among aged persons:analysis of SAGE selected countries [J].BMC Geriatr,2023,23(1):145.
[2]SALEH G A,BATOUTY N M,HAGGAG S,et al.The role of medical image modalities and AI in the early detection,diagnosis and grading of retinal diseases:a survey[J].Bioengineering,2022,9(8):366.
[3]WANG S,CHEN Y,YI Z.A Multi-Scale Attention Fusion Network for Retinal Vessel Segmentation[J].Applied Sciences,2024,14(7):2955.
[4]CHEN C,CHUAH J H,ALI R,et al.Retinal Vessel Segmentation Using Deep Learning:A Review [J].IEEE Access,2021,9:111985-112004.
[5]XU M,CHEN S,GAO X,et al.Research on Fast Multi-Threshold Image Segmentation Technique Using Histogram Analysis[J].Electronics,2023,12(21):4446.
[6]WISAENG K.Retinal blood vessel segmentation using density-based fuzzy C-means clustering and vessel neighborhood connected component[J].Measurement,2024,242:116229.
[7]KAR S S,MAITY S P.Blood vessel extraction and optic disc removal using curvelet transform and kernel fuzzy c-means [J].Computer in Biology and Medicine,2016,70:174-189.
[8]RICCI E,PERFETTI R.Retinal blood vessel segmentation using line operators and support vector classification [J].IEEE Transactions on Medical Imaging,2007,26(10):1357-1365.
[9]SKOUTA A,ELMOUFIDI A,JAI-ANDALOUSSI S,et al.Semantic Segmentation of Retinal Blood Vessels from Fundus Images by using CNN and the Random Forest Algorithm[C]//SENSORNETS.2022:163-170.
[10]KUFEL J,BARGIEL-LACZEK K,KOCOT S,et al.What ismachine learning,artificial neural networks and deep learning?－Examples of practical applications in medicine[J].Diagnostics,2023,13(15):2582.
[11]COŞKUN M,YILDIRIM Ö,UÇAR A,et al.An overview of popular deep learning methods[J].European Journal of Technique,2017,7(2):165-176.
[12]KRIZHEVSKY A,SUTSKEVER I,HINTON G E.Imagenetclassification with deep convolutional neural networks[C]//Advances in Neural Information Processing Systems.2012.
[13]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.
[14]RONNEBERGER O,FISCHER P,BROX T.U-net:Convolutionalnetworks for biomedical image segmentation[C]//Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015:18th International Conference.Springer,2015:234-241.
[15]OKTAY O.Attention u-net:Learning where to look for the pancreas[J].arXiv:1804.03999,2018.
[16]YANG X,LI Z,GUO Y,et al.DCU-net:A deformable convolutional neural network based on cascade U-net for retinal vessel segmentation[J].Multimedia Tools and Applications,2022,81(11):15593-15607.
[17]SZEGEDY C,LIU W,JIA Y,et al.Going deeper with convolutions[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2015:1-9.
[18]SZEGEDY C,VANHOUCKE V,IOFFE S,et al.Rethinking the inception architecture for computer vision[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2016:2818-2826.
[19]ZHOU Z,RAHMAN SIDDIQUEE M M,TAJBAKHSH N,et al.Unet++:A nested u-net architecture for medical image segmentation[C]//Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support:4th International Workshop,DLMIA 2018,and 8th International Workshop,ML－CDS 2018,Held in Conjunction with MICCAI 2018.Springer,2018:3-11.
[20]LIU Y P,RUI X,LI Z,et al.Feature pyramid U-Net for retinal vessel segmentation[J].IET Image Processing,2021,15(8):1733-1744.
[21]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.
[22]SHI Z,WANG T,HUANG Z,et al.MD-Net:A multi-scaledense network for retinal vessel segmentation[J].Biomedical Signal Processing and Control,2021,70:102977.
[23]LI J,GAO G,YANG L,et al.A retinal vessel segmentation network with multiple-dimension attention and adaptive feature fusion[J].Computers in Biology and Medicine,2024,172:108315.
[24]GUO C,SZEMENYEI M,YI Y,et al.Sa-unet:Spatial attention u-net for retinal vessel segmentation[C]//2020 25th International Conference on Pattern Recognition(ICPR).IEEE,2021:1236-1242.
[25]GHIASI G,LIN T Y,LE Q V.Dropblock:A regularizationmethod for convolutional networks[C]//Advances in Neural Information Processing Systems.2018.
[26]GOLIAŠ M,ŠIKUDOVÁ E.Retinal blood vessel segmentationand inpainting networks with multi-level self-attention[J].Biomedical Signal Processing and Control,2025,102:107343.
[27]SHEN X,XU J,JIA H,et al.Self-attentional microvessel segmentation via squeeze-excitation transformer Unet[J].Compu-terized Medical Imaging and Graphics,2022,97:102055.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed