摘要
视网膜血管形态结构是反映人体健康的重要指标,针对现有视网膜血管分割存在主血管模糊、微细血管断裂和视盘误分割等问题,提出多尺度特征融合双U型视网膜分割算法。首先,利用低层U-Net高效循环残差模块对眼底图像进行粗粒度分割,得到视网膜血管初步轮廓。其次,将粗分割图与原始特征图像素相乘送入高层U-Net,利用其缩放宽残差模块进行细粒度图像解码,丰富视网膜血管细节信息。同时利用3路径注意力机制复合性连接双网络的编码层与解码层,实现特征映射跨网络传播,减小上下文语义差异。最后,融合双层网络输出提取血管区域,双U型网络能够更深层次提取血管像素,精准分割出视网膜细节。在DRIVE与STARE数据集上进行实验,其准确率分别为96.45%和97.02%,敏感度分别为83.35%和81.40%,特异性分别为98.38%和98.83%,总体性能优于现有算法。
The morphological structure of retinal vessels is an important index to reflect human health.In order to solve the existing problems in retinal vessel segmentation,such as blurred main vessels,broken microvessels and false segmentation of optic disc,the multi-scale feature fusion double U-shaped retinal segmentation algorithm is proposed.Firstly,the low level U-Net efficient cyclic residual module is used for coarse-grained segmentation of fundus images to obtain the initial contour of retinal vessels.Secondly,the coarse segmentation image is multiplied by the pixels of the original feature image into the high level U-Net,and the scaling wide residual model is used to decode the fine-grained image to enrich the details of retinal vessels.At the sametime,the three pathway attention mechanism is used to connect the encoding layer and decoding layer of the double network in a compound way to realize the cross network propagation of feature mapping and reduce the semantic difference of context.Finally,the double Ushaped network can extract vascular pixels at a deeper level and accurately segment retinal details.Experiments were conducted on DRIVE and STARE datasets,the accuracy was 96.45%and 97.02%,the sensitivity was 83.35%and 81.40%,and the specificity was 98.38%and 98.83%,respectively.The overall performance is better than existing algorithms.
作者
梁礼明
周珑颂
余洁
陈鑫
LIANG Liming;ZHOU Longsong;YU Jie;CHEN Xin(School of Electrical Engineering and Automation,Jiangxi University of Science and Technology,Ganzhou,Jiangxi 341000,China)
出处
《光电子．激光》
CAS
CSCD
北大核心
2022年第3期272-282,共11页
Journal of Optoelectronics·Laser
基金
国家自然科学基金(51365017,61463018)
江西省自然科学基金面上项目(20192BAB205084)
江西省教育厅科学技术研究重点项目(GJJ170491)资助项目。
关键词
视网膜血管
双U型网络
循环残差
多尺度特征融合
三路径注意力机制
retinal vessel
double U-shaped
circular residual
multi-scale feature fusion
three-pathway attention mechanism
