[1]许冬,李浩,周利晓,等.基于卷积神经网络UNet构建糖尿病性黄斑水肿自动识别模型[J].眼科新进展,2020,40(4):357-361.[doi:10.13389/j.cnki.rao.2020.0082]
　XU Dong,LI Hao,ZHOU Lixiao,et al.Model of automatic identification of diabetic macular edema via convolutional neural networks UNet[J].Recent Advances in Ophthalmology,2020,40(4):357-361.[doi:10.13389/j.cnki.rao.2020.0082]
点击复制

基于卷积神经网络UNet构建糖尿病性黄斑水肿自动识别模型/HTML
分享到：

《眼科新进展》[ISSN:1003-5141/CN:41-1105/R]

卷:
40卷
期数:
2020年4期
页码:
357-361
栏目:
应用研究
出版日期:
2020-04-05

文章信息/Info

Title:
Model of automatic identification of diabetic macular edema via convolutional neural networks UNet
作者:
许冬; 李浩; 周利晓; 吕梁
450052　河南省郑州市，郑州大学第五附属医院眼科（许冬，周利晓，吕梁）；611731　四川省成都市，电子科技大学（李浩）
Author(s):
XU Dong1; LI Hao2; ZHOU Lixiao1; Lü Liang1
1.Department of Ophthalmology,the Fifth Affiliated Hospital of Zhengzhou University,Zhengzhou 450052,Henan Province,China
2.University of Electronic Science and Technology of China,Chengdu 611731,Sichuan Province,China
关键词:
糖尿病性视网膜病变; 黄斑水肿; 深度学习; 卷积神经网络; 光学相干断层扫描
Keywords:
diabetic retinopathy;  macular edema;  deep learning;  convolutional neural networks;  optical coherence tomography
分类号:
R770.4
DOI:
10.13389/j.cnki.rao.2020.0082
文献标志码:
A
摘要:
目的　通过卷积神经网络UNet构建光学相干断层扫描（optical coherence tomography，OCT）图像中糖尿病性黄斑水肿的自动识别模型，并通过相关指标判断其价值。方法　利用开源的OCT数据集2014_BOE_Srinivasan和OCT2017训练卷积神经网络UNet模型，并结合我院2018年1月至2019年5月的60例糖尿病性黄斑水肿患者的OCT检查影像结果共同组成数据集来验证模型。最后通过该模型的损失函数变化和精确度变化，以及绘制受试者工作特征曲线来评价模型。结果　卷积神经网络UNet对单张图像的处理时间在75 ms左右。且损失函数变化图显示当模型训练到一定程度后，损失数值逐渐趋于收敛。验证集的精确度变化图显示精确度可以达到0.9左右，并且随着训练次数的不断增加，精确度逐渐趋于稳定。最后根据测试结果绘制了受试者工作特征曲线，其曲线下面积达到0.902，提示该模型具有较高诊断能力。结论　利用卷积神经网络UNet可以准确快速地分割出糖尿病性黄斑水肿区域，有望辅助临床医师的诊断与治疗。
Abstract:
Objective　To establish an automatic model of diabetic macular edema (DME) in optical coherence tomography (OCT) images by convolutional neural networks UNet, and determine its value through relevant indicators.Methods　The convolutional neural networks UNet model trained by using the two open sources OCT data sets 2014_BOE_Srinivasan and COT2017 in combination with the OCT images of 60 DME patients from January 2018 to May 2019 was used to validate the model. The model is evaluated by loss curve, validation accuracy curve and receiver operating characteristic (ROC) curve.Results　For a single image to be segmented, the processing time was only about 75 ms. The loss curve figure showed that when the model was trained to a certain level, the loss function value tended to converge. The validation accuracy figure showed that the validation accuracy of OCT data can reach 0.9 or above, and the accuracy gradually tended to be stable as the number of training iterations increased continuously. In addition, ROC curve was drawn based on test results, and the area under curve (AUC) of ROC reached 0.902, which indicated a high diagnostic capacity.Conclusion　Convolutional neural networks UNet can accurately and quickly segment the DME area, which may assist the diagnosis or treatment of clinicians.

参考文献/References:

［1］　张美霞,杨兰芬,罗成仁,严密,孟丹,沈齐,等.糖尿病视网膜病变黄斑水肿的临床分析［J］.中华眼底病杂志,2003,19(2):83-86.
ZHANG M X,YANG L F,LUO C R,YAN M,MENG D,SHEN Q,et al.Clinical analysis of the results of macular edema in diabetic retinopathy［J］.Chin J Ocul Fundus Dis,2003,19(2):83-86.
［2］　YANG W Y,LU J M,WENG J P,JIA W P,JI L N,XIAO J Z,et al.Prevalence of diabetes among men and women in China［J］.N Engl J Med,2010,362(12):1090-1101.
［3］　FANG L,CUNEFARE D,WANG C,GUYMER R H,LI S,FARSIU S.Automatic segmentation of nine retinal layer boundaries in OCT images of non-exudative AMD patients using deep learning and graph search［J］.Biomed Opt Express,2017,8(5):2732-2744.
［4］　SCHLEGL T,WALDSTEIN S M,BOGUNOVIC H,ENDSTRAβER F,SADEGHIPOUR A,PHILIP A M,et al.Fully Automated detection and quantification of macular fluid in OCT using deep learning［J］.Ophthalmology,2018,125(4):549-558.
［5］　GARGEYA R,LENG T.Automated identification of diabetic retinopathy using deep learning［J］.Ophthalmology,2017,124(7):962-969.
［6］　LEE C S,BAUGHMAN D M,LEE A Y.Deep learning is effective for the classification of OCT images of normal versus age-related macular degeneration［J］.Ophthalmol Retina,2017,1(4):322-327.
［7］　ASAOKA R,MURATA H,IWASE A,ARAIE M.Detecting preperimetric glaucoma with standard automated perimetry using a deep learning classifier［J］.Ophthalmology,2016,123(9):1974-1980.
［8］　RONNEBERGER O,FISCHER P,BROX T.U-net:Convolutional networks for biomedical image segmentation［C］.Miccai:Springer,2015.
［9］　SRINIVASAN P P,KIM L A,METTU P S,COUSINS S W,COMER G M,IZATT J A,et al.Fully automated detection of diabetic macular edema and dry age-related macular degeneration from optical coherence tomography images［J］.Biomed Opt Express,2014,5(10):3568-3577.
［10］　KERMANY D S,GOLDBAUM M,CAI W,VALENTIM C C S,LIANG H Y,BAXTER S L,et al.Identifying medical diagnoses and treatable diseases by image-based deep learning［J］.Cell,2018,172(5):1122-1131.
［11］　SHELHAMER E,LONG J,DARRELL T.Fully convolutional networks for semantic segmentation［J］.IEEE Trans Pattern Anal Mach Intell,2017,39(4):640-651.
［12］　ZEILER M D,FERGUS R.Visualizing and understanding convolutional networks［C］.ECCV:Springer,2014.
［13］　WILKINS G R,HOUGHTON O M,OLDENBURG A L.Automated segmentation of intraretinal cystoid fluid in optical coherence tomography［J］.IEEE Trans Biomed Eng,2012,59(4):1109-1114.
［14］　GARVIN M K,ABRAMOFF M D,KARDON R,RUSSELL S R,WU X,SONKA M.Intraretinal layer segmentation of macular optical coherence tomography images using optimal 3-D graph search［J］.IEEE Trans Med Imaging,2008,27(10):1495-1505.
［15］　CHEN X J,NIEMEIJER M,ZHANG L,LEE K,ABRAMOFF M D,SONKA M.Three-dimensional segmentation of fluid-associated abnormalities in retinal OCT:probability constrained graph-search-graph-cut［J］.IEEE Trans Med Imaging,2012,31(8):1521-1531.
［16］　WU J,PHILIP A M,PODKOWINSKI D,GERENDAS B S,LANGS G,SIMADER C,et al.Multivendor spectral-domain optical coherence tomography dataset,observer annotation performance evaluation,and standardized evaluation framework for intraretinal cystoid fluid segmentation［J］.J Ophthalmol,2016,2016:3898750.
［17］　LIN T Y,DOLLAR P,GIRSHICK R,HE K,HARIHARAN B,BELONGIE S.Feature pyramid networks for object detection［C］.CVPR:Springer,2017.

相似文献/References:

[1]张 洁 孟小妹 张亦农.重度非增殖性与增殖性糖尿病性视网膜病变多焦视网膜电图一阶函数核的特征分析[J].眼科新进展,2008,28(9):000.
[2]李国兴 韩丽英.曲安奈德玻璃体腔注射治疗黄斑水肿[J].眼科新进展,2008,28(9):000.
[3]周正申 孙静芬.白内障超声乳化术后黄斑区光学相干断层扫描观察[J].眼科新进展,2013,33(2):000.
[4]刘广峰 蔺洁 范颖 顾铮 高新晓 朱岩 孟忻 王华 贾宇颂 付秀稳 汪军.非增生型糖尿病视网膜病变黄斑水肿患者黄斑区谱域光学相干断层扫描中高反射信号的特征分析[J].眼科新进展,2013,33(4):000.
[5]鲁丹 王康 王艳玲 林鹏耀 李茜 李明铭.后Tenon囊下注射曲安奈德辅助全视网膜光凝治疗糖尿病视网膜病变[J].眼科新进展,2013,33(5):000.
[6]张祺 冯劼 李培凤 金翼.Lucentis联合氪黄激光治疗视网膜分支静脉阻塞黄斑水肿的疗效观察[J].眼科新进展,2013,33(12):000.
[7]陆秉文 吴星伟.视网膜静脉阻塞黄斑水肿患者玻璃体内注射ranibi-zumab后视野变化[J].眼科新进展,2013,33(12):000.
[8]徐斌 戴奕娟 梁丽 杨露. 传统格栅样激光与577nm黄色微脉冲激光光凝治疗黄斑水肿疗效对比[J].眼科新进展,2014,34(2):181.
[9]陈素芳 曾莉 邵毅. 非诺贝特在糖尿病视网膜病变中的应用进展[J].眼科新进展,2014,34(3):283.
[10]陈娟,吕红彬. 抗VEGF药物在糖尿病性视网膜病变治疗中的应用[J].眼科新进展,2014,34(4):397.[doi:10.13389/j.cnki.rao.2014.0110]

备注/Memo

备注/Memo:
河南省医学科技攻关项目（编号：201503131）
更新日期/Last Update: 2020-04-05