摘要
视网膜血管检测在眼底疾病的诊断和治疗中具有重要的临床价值。但是,由于眼底图像特征的复杂性和多样性,大部分的视网膜分割方法存在血管分割性能低、抗噪声干扰能力弱和对病灶敏感等问题,为此,提出了一种集成深度分类神经网络对像素点分类的方法。首先利用不同的残差网络模型来分类像素点,获得血管分割图像;然后通过集成学习的方法对各个模型的分割结果进行处理,获得最终的视网膜血管分割图像。在STARE、DRIVE和CHASE数据集上的实验仿真结果显示,分割准确率分别达到97.36%,95.57%,96.36%,特异性分别达到98.06%,97.76%,97.84%,F-measure分别达到84.98%,82.25%,79.87%。比R2U_Net的F-measure分别提高了0.23%,0.54%,0.59%。
Retinal blood vessel detection has important clinical value in the diagnosis and treatment of fundus diseases.However,due to the complexity and diversity of fundus image features,most retinal segmentation methods have some problems such as low performance of blood vessel segmentation,weak anti-noise interference,and sensitivity to lesions.Therefore,a pixel points classification method based on ensembled classified deep neural network is proposed.Firstly,different residual network models are used to classify pixel points and get the vascular segmentation image.Secondly,through the ensemble learning method,the segmentation results of each model are processed to obtain the final retinal vascular segmentation image.The simulation results on STARE,DRIVE,and CHASE datasets show that the segmentation accuracy is 97.36%,95.57%,96.36%,the specificity is 98.06%,97.76%,97.84%,and the F-measure is 84.98%,82.25%,79.87%.The F-measure is 0.23%,0.54%,and 0.59%higher than R2U_Net.
作者
蒋芸
王发林
张海
JIANG Yun;WANG Fa-lin;ZHANG Hai(College of Computer Science and Engineering,Northwest Normal University,Lanzhou 730070,China)
出处
《计算机工程与科学》
CSCD
北大核心
2021年第5期862-871,共10页
Computer Engineering & Science
基金
国家自然科学基金(61962054,61163036)
2016年甘肃省科技计划资助自然科学基金(1606RJZA047)。
关键词
深度学习
卷积神经网络
图像分割
集成学习
deep learning
convolutional neural network
image segmentation
ensemble learning
作者简介
蒋芸(1970-),女,浙江绍兴人,博士,教授,研究方向为数据挖掘、粗糙集理论及应用,E-mail:jiangyun@nwnu.edu.cn,通信地址:730070甘肃省兰州市西北师范大学计算机科学与工程学院;王发林(1996-),男,甘肃庄浪人,硕士生,CCF会员(A5303G),研究方向为数据挖掘,E-mail:1753190368@qq.com;张海(1995-),男,江西赣州人,硕士生,CCF会员(91445G),研究方向为数据挖掘,E-mail:Haiccheung1995@gmail.com。
