基于改进YOLOv5的眼底出血点检测算法

doi:10.16088/j.issn.1001-6600.2023053101

摘要/Abstract

摘要： 糖尿病视网膜病变眼底图像中出血点病灶尺寸小且分布密集,导致现有算法难以实现该病灶的精确检测及定位。本文提出一种RCA-YOLO出血病灶检测算法,算法在YOLOv5s基础上,首先在主干网络中使用RCA-Net模块,使网络在获得各通道间联系的同时保留目标病灶的位置信息,增强网络对出血区域的特征提取及定位能力;然后在特征融合阶段采用轻量化特征金字塔网络Tiny-BiFPN,在减少网络参数量的同时,实现高效率的多尺度特征融合;最后提出小目标特征增强模块,提升算法对小出血点病灶的检测精度。实验结果表明,改进后的RCA-YOLO算法能够准确地检测并定位出血点病灶,平均检测准确率(mAP)可达79.3%,较YOLOv5s算法提高了9.5个百分点,其检测结果同样优于Faster R-CNN、YOLOv6s、YOLOv7和YOLOv8s等主流算法。

关键词: 眼底图像, 糖尿病视网膜病变, 目标检测, 出血病灶, 特征融合

Abstract: The small size and dense distribution of bleeding point lesions in the fundus image of diabetic retinopathy make it difficult for the existing algorithms to achieve accurate detection and localization of the lesions. A RCA-YOLO bleeding lesion detection algorithm is proposed. Based on YOLOv5s, the RCA-Net module is first used in the backbone network, so that the network can obtain the connection between each channel while retaining the location information of the target lesion, and enhance the feature extraction and localization ability of the network for the bleeding area. In the feature fusion stage, the lightweight feature pyramid network Tiny-BiFPN is adopted to reduce the number of network parameters and achieve high-efficiency multi-scale feature fusion. Finally, a small target feature enhancement module is proposed to improve the detection accuracy of the algorithm for small bleeding point lesions. The experimental results show that the improved RCA-YOLO algorithm can accurately detect and locate bleeding point lesions, and the average detection accuracy (mAP) can reach 79.3%, which is 9.5 percentage points higher than that of YOLOv5s algorithm, and its detection results are also better than mainstream algorithms such as Faster R-CNN, YOLOv6s, YOLOv7 and YOLOv8s.

Key words: fundus image, diabetic retinopathy, object detection, hemorrhagic lesion, feature fusion

中图分类号: R587.2;R774.1;TP183;TP391.41

吕辉, 吕卫峰. 基于改进YOLOv5的眼底出血点检测算法[J]. 广西师范大学学报（自然科学版）, 2024, 42(3): 99-107.

LÜ Hui, LÜ Weifeng. Fundus Hemorrhagic Spot Detection Algorithm Based on Improved YOLOv5[J]. Journal of Guangxi Normal University(Natural Science Edition), 2024, 42(3): 99-107.

参考文献

[1] WONG T Y, SUN J, KAWASAKI R, et al. Guidelines on diabetic eye care: the International Council of Ophthalmology recommendations for screening, follow-up, referral, and treatment based on resource settings[J]. Ophthalmology, 2018, 125(10): 1608-1622. DOI: 10.1016/j.ophtha.2018.04.007.
[2] ZHAO Y B, FANG F Z. Measurement of the peripheral aberrations of human eyes: a comprehensive review[J]. Nanotechnology and Precision Engineering, 2020, 3(2): 53-68. DOI: 10.1016/j.npe.2020.05.001.
[3] 张琳.糖尿病视网膜病变早期筛防的意义[J].中国医药指南,2019,17(2):80. DOI: 10.15912/j.cnki.gocm.2019.02.069.
[4] 卢旭,郑世宝,彭昊玥.基于卷积神经网络的糖尿病视网膜眼底图像出血病变检测[J].电视技术,2019,43(1):84-89,110. DOI: 10.16280/j.videoe.2019.01.016.
[5] 林秀琴,熊义斌,肖键,等.免散瞳眼底照相在糖尿病视网膜病变筛查中的应用分析[J].国际眼科杂志,2019,19(1):135-138. DOI: 10.3980/j.issn.1672-5123.2019.1.31.
[6] 李子鹏.眼底荧光造影检查在糖尿病眼底病变筛查中的价值分析[J].中国实用医药,2022,17(15):78-80. DOI: 10.14163/j.cnki.11-5547/r.2022.15.023.
[7] GARCÍA M, LÓPEZ M I, ALVAREZ D, et al. Assessment of four neural network based classifiers to automatically detect red lesions in retinal images[J]. Medical Engineering & Physics, 2010, 32(10): 1085-1093. DOI: 10.1016/j.medengphy.2010.07.014.
[8] KAUR N, CHATTERJEE S, ACHARYYA M, et al. A supervised approach for automated detection of hemorrhages in retinal fundus images[C]// 2016 5th International Conference on Wireless Networks and Embedded Systems (WECON). Los Alamitos, CA: IEEE Computer Society, 2016: 1-5. DOI: 10.1109/WECON.2016.7993461.
[9] KANDE G B, SAVITHRI T S, SUBBAIAH P V, et al. Detection of red lesions in digital fundus images[C]// 2009 IEEE International Symposium on Biomedical Imaging: from Nano to Macro. Los Alamitos, CA: IEEE Computer Society, 2009: 558-561. DOI: 10.1109/ISBI.2009.5193108.
[10] 肖志涛,赵北方,张芳,等.基于k均值聚类和自适应模板匹配的眼底出血点检测方法[J].中国生物医学工程学报,2015,34(3):264-271. DOI: 10.3969/j.issn.0258-8021.2015.03.002.
[11] SHENAVARMASOULEH F, ARABNIA H R. DRDr: automatic masking of exudates and microaneurysms caused by diabetic retinopathy using mask R-CNN and transfer learning[C]// Advances in Computer Vision and Computational Biology. Cham: Springer, 2021: 307-318. DOI: 10.1007/978-3-030-71051-4_24.
[12] AKUT R R. FILM: finding the location of microaneurysms on the retina[J]. Biomedical Engineering Letters, 2019, 9(4): 497-506. DOI: 10.1007/s13534-019-00136-6.
[13] 高玮玮,单明陶,宋楠,等.嵌入SENet的改进YOLOv4眼底图像微动脉瘤自动检测算法[J].生物医学工程学杂志,2022,39(4):713-720. DOI: 10.7507/1001-5515.202203022.
[14] 高玮玮,杨亦乐,方宇,等.多特征尺度融合改进Faster-RCNN视网膜微动脉瘤自动检测算法[J].光子学报,2023,52(4):0410002. DOI: 10.3788/gzxb20235204.0410002.
[15] 侯高峰,房丰洲.基于深度学习的糖尿病眼底病变检测研究[J].激光与光电子学进展,2023,60(2):0217001. DOI: 10.3788/LOP212505.
[16] 张焕龙,齐企业,张杰,等.基于改进YOLOv5的输电线路鸟巢检测方法研究[J].电力系统保护与控制,2023,51(2):151-159. DOI: 10.19783/j.cnki.pspc.220428.
[17] 黄叶祺,王明伟,闫瑞,等.基于改进的YOLOv5金刚石线表面质量检测[J].广西师范大学学报(自然科学版),2023,41(4):123-134. DOI: 10.16088/j.issn.1001-6600.2022112106.
[18] HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Los Alamitos, CA: IEEE Computer Society, 2018: 7132-7141. DOI: 10.1109/CVPR.2018.00745.
[19] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]// Computer Vision-ECCV 2018. Cham: Springer, 2018: 3-19. DOI: 10.1007/978-3-030-01234-2_1.
[20] TAN M X, PANG R M, LE Q V. EfficientDet: scalable and efficient object detection[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2020: 10778-10787. DOI: 10.1109/CVPR42600.2020.01079.
[21] LI T, GAO Y Q, WANG K, et al. Diagnostic assessment of deep learning algorithms for diabetic retinopathy screening[J]. Information Sciences, 2019, 501: 511-522. DOI: 10.1016/j.ins.2019.06.011.
[22] LI C Y, LI L L, JIANG H L, et al. YOLOv6: a single-stage object detection framework for industrial applications[EB/OL]. (2022-09-07)[2023-06-01]. https://arxiv.org/abs/2209.02976. DOI: 10.48550/arXiv.2209.02976.
[23] WANG C Y, BOCHKOVSKIY A, LIAO H Y M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2023: 7464-7475. DOI: 10.1109/CVPR52729.2023.00721.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed