基于ResNeSt网络的糖网病及糖尿病黄斑水肿联合分级

doi:10.3969/j.issn.0255-8297.2025.05.011

摘要/Abstract

摘要： 糖网病（diabetic retinopathy,DR）和糖尿病黄斑水肿（diabeticmacular edema,DME）是造成人类失明的主要原因。本文基于ResNeSt网络关注DR和DME之间的密切关联以实现两者的联合分级，从而提高分级精度。使用ResNeSt网络的不同层级分别提取DR和DME的共有特征及两者的特有特征，在此基础上利用互关注模块进行特征融合，实现两者的各自分级。在Messidor数据集上，同时使用ResNeSt网络和互关注模块时DR和DME的分级准确率分别为95.6%和95.0%，联合准确率为86.7%；而仅使用ResNeSt网络时两者的分级准确率分别为95.0%和90.8%。在IDRiD数据集上两者的联合准确率为66.3%。数据集的研究结果表明，基于ResNeSt网络对DR和DME进行联合分级可以提高分级精度。

关键词: 糖网病, 黄斑水肿, ResNeSt, 注意力机制, 联合分级

Abstract: Diabetic retinopathy (DR) and diabetic macular edema (DME) are the major causes of blindness in humans. This study proposes to leverage the ResNeSt network to focus on the close relationship between DR and DME, and use it to achieve their joint grading to improve grading accuracy. Specifically, common features and unique features of DR and DME are extracted by using different stages of ResNeSt. Subsequently, a mutual attention module for feature fusion is employed to realize their respective grading. On the Messidor dataset, by using both ResNeSt network and the mutual attention module, the grading accuracies of DR and DME were 95.6% and 95.0%, respectively, and the joint accuracy was 86.7%. In contrast, when only the ResNeSt network was used, the grading accuracies of the two were 95.0% and 90.8%, respectively. On the IDRiD dataset, the joint accuracy of the two reached 66.3%. The results on the datasets indicate that the proposed joint grading of DR and DME based on the ResNeSt network can improve grading accuracy.

Key words: diabetic retinopathy, macular edema, ResNeSt, attention mechanism, joint grading

中图分类号:

TP391.4

张爱玲, 杨林英, 闫士举. 基于ResNeSt网络的糖网病及糖尿病黄斑水肿联合分级[J]. 应用科学学报, 2025, 43(5): 849-862.

ZHANG Ailing, YANG Linying, YAN Shiju. Joint Grading of Diabetic Retinopathy and Diabetic Macular Edema Based on ResNeSt Network[J]. Journal of Applied Sciences, 2025, 43(5): 849-862.

参考文献

[1] 陈瑜倩, 吕东辉, 宋安平, 等. 基于深度学习的糖尿病足伤口TEXAS分期研究[J]. 应用科学学报, 2024, 42(3): 437-446. Chen Y Q, Lyu D H, Song A P, et al. TEXAS staging of diabetic foot wounds based on deep learning approach [J]. Journal of Applied Sciences, 2024, 42(3): 437-446. (in Chinese)
[2] 刘利丽. 糖尿病性黄斑水肿与患者外周血全血细胞计数的相关性研究[D]. 南充: 川北医学院, 2023.
[3] Wang Y L, Yu M, Hu B J, et al. Deep learning-based detection and stage grading for optimising diagnosis of diabetic retinopathy [J]. Diabetes-Metabolism Research and Reviews, 2021, 37(4): e3445.
[4] 梁礼明, 董信, 何安军, 等. 融合注意力线性特征多样化的DR分级模型[J]. 光电子· 激光, 2024, 35(6): 612-622. Liang L L, Dong X, He A J, et al. DR grading model of fusing attention linear feature diversification [J]. Optoelectronics · Laser, 2024, 35(6): 612-622. (in Chinese)
[5] Cheena M, Sakuntala M, Biswaranjan A, et al. Using deep learning architectures for detection and classification of diabetic retinopathy [J]. Sensors, 2023, 23(12): 5726-5743.
[6] Abbas Q, Daadaa Y, Rashid U, et al. HDR-EfficientNet: a classification of hypertensive and diabetic retinopathy using optimize efficientnet architecture [J]. Diagnostics, 2023, 13(20): 3236-3265.
[7] Hemanth D J, Deperlioglu O, Kose U. An enhanced diabetic retinopathy detection and classification approach using deep convolutional neural network [J]. Neural Computing and Applications, 2019, 32(3): 707-721.
[8] Jayakumari C, Lavanya V, Sumesh E P. Automated diabetic retinopathy detection and classification using ImageNet convolution neural network using fundus images [C]// International Conference on Smart Electronics and Communication, 2020: 577-582.
[9] Wang C Y, Mukundan A, Liu Y S, et al. Optical identification of diabetic retinopathy using hyperspectral imaging [J]. Journal of Personalized Medicine, 2023, 13(6): 939-950.
[10] Jaichandran R, Sivasubramanian V, Jayaprakash. Detection of diabetic retinopathy using convolutional neural networks [J]. ECS Transactions, 2022, 107: 13321-13328.
[11] Gayathri S, Gopi V P, Palanisamy P. A lightweight CNN for diabetic retinopathy classification from fundus images [J]. Biomedical Signal Processing and Control, 2020, 62: 102-115.
[12] 郑智文, 甘健侯, 周菊香, 等. 基于注意力网络推理图的细粒度图像分类[J]. 应用科学学报, 2022, 40(1): 36-46. Zheng Z W, Gan J H, Zhou J X, et al. Fine-grained image classification based on inference graph of attention network [J]. Journal of Applied Sciences, 2022, 40(1): 36-46. (in Chinese)
[13] 雷前慧, 潘丽丽, 邵伟志, 等. 基于三重注意力机制的新冠肺炎病灶分割模型[J]. 应用科学学报, 2022, 40(1): 105-115. Lei Q H, Pan L L, Shao W Z, et al. Segmentation model of covid-19 lesions based on triple attention mechanism [J]. Journal of Applied Sciences, 2022, 40(1): 105-115. (in Chinese)
[14] Zhang H, Wu C R, Zhang Z Y, et al. ResNeSt: split-attention networks [C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022: 2735-2745.
[15] 蔡乾宏. 基于深度学习的眼底图像分析及糖网病病变检测技术研究[D]. 贵阳: 贵州大学, 2021.
[16] Li X M, Hu X W, Yu L Q, et al. CANet: cross-disease attention network for joint diabetic retinopathy and diabetic macular edema grading [J]. IEEE Transactions on Medical Imaging, 2020, 39(5): 1483-1493.
[17] Decenciere E, Zhang X, Cazuguel G, et al. Feedback on a publicly distributed image database: the messidor database [J]. Image Analysis & Stereology, 2014, 33(3): 231-234
[18] Porwal P, Pachade S, Kamble R, et al. Indian diabetic retinopathy image dataset (IDRiD): a database for diabetic retinopathy screening research [J]. Data, 2018, 3(3): 1-8.
[19] Sanchez C I, Niemeijer M, Dumitrescu A V, et al. Evaluation of a computer-aided diagnosis system for diabetic retinopathy screening on public data [J]. Investigative Ophthalmology & Visual Science, 2011, 52(7): 4866-4871.
[20] Hu J, Shen L, Albanie S. Squeeze-and-excitation networks [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 42(8): 2011-2023.
[21] Li X, Wang W, Hu X. Selective kernel networks [C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 510-519.
[22] Xie S, Girshick R, Doll'ar P, et al. Aggregated residual transformations for deep neural networks [C]//30TH IEEE Conference on Computer Vision and Pattern Recognition, 2017: 5987-5995.
[23] Pires R, Avila S, Jelinek H F, et al. Beyond lesion-based diabetic retinopathy: a direct approach for referral [J]. IEEE Journal of Biomedical and Health Informatics, 2017, 21(1): 193-200.
[24] Vo H H, Verma A. New deep neural nets for fine-grained diabetic retinopathy recognition on hybrid color space [C]//18th IEEE International Symposium on Multimedia, 2016: 209-215.
[25] Seoud L, Hurtut T, Chelbi J, et al. Red lesion detection using dynamic shape features for diabetic retinopathy screening [J]. IEEE Transactions on Medical Imaging, 2016, 35(4): 1116-1126.
[26] Al-Bander B, Al-Nuaimy W, Al-Taee M A, et al. Diabetic macular edema grading based on deep neural networks [C]//Third International Workshop on Ophthalmic Medical Image Analysis, 2016: 121-128.