人工智能结合光相干断层扫描技术辅助识别5种眼底病变系统的应用和价值评估  

The application and value evaluation of assisted diagnosis system for five fundus lesion based on artificial intelligence combined with optical coherence tomography

作　　者：马健 陈淑梅 王敏 吴福理 吴健 方肖云[1] Ma Jian;Chen Shumei;Wang Min;Wu Fuli;Wu Jian;Fang Xiaoyun(Department of Ophthalmology,Eye Center of the Second Affiliated Hospitial,School of Medicine,Zhejiang University,Hangzhou 310009,China;Department of Ophthalmology,The First People’s Hospital of Jiashan County,Jiaxing 314100,China;Department of Ophthalmology,Nanxun District People's Hospital,Huzhou 313009,China;Ruiyi Artificial Intelligence Research Center,Zhejiang University,Hangzhou 310000,China;School of Public Health,Zhejiang University Medical College,Hangzhou 310000,China)

机构地区：[1]浙江大学医学院附属第二医院眼科中心,杭州310009 [2]浙江省嘉兴市嘉善县第一人民医院眼科,嘉兴314100 [3]浙江省湖州市南浔区人民医院眼科,湖州313009 [4]浙大睿医人工智能研究中心,杭州310000 [5]浙江大学医学院公共卫生学院,杭州310000

出　　处：《中华眼底病杂志》2022年第2期126-131,共6页Chinese Journal of Ocular Fundus Diseases

基　　金：浙江省自然科学基金(LY21H120002)。

摘　　要：目的建立基于深度学习光相干断层扫描(OCT)图像眼底病变的眼底智能辅助诊断系统,初步评估其应用价值。方法诊断性试验研究。2016年至2019年期间于浙江大学医学院附属第二医院眼科中心就诊的25000例患者的25000张OCT图像作为眼底智能辅助诊断系统的训练集和验证集。其中,黄斑前膜、黄斑水肿、黄斑裂孔、脉络膜新生血管(CNV)、老年性黄斑变性(AMD)各5000张。训练集、验证集分别为18124、6876张。通过迁移学习Attention ResNet结构算法,对OCT图像进行特征性病变识别,通过特定程序提取疾病特征,根据目标病变的统计特征,将给定的图像与其他类型的疾病进行区分。初步形成黄斑前膜、黄斑水肿、黄斑裂孔、CNV、AMD的模型算法,建立5种模型的眼底智能辅助诊断系统。应用受试者工作特征曲线及曲线下面积(AUC)、灵敏度、特异性评估眼底智能辅助诊断系统中各模型辅助诊断的性能。结果眼底智能辅助诊断系统中,黄斑前膜模型辅助诊断的灵敏度、特异性分别为93.5%、99.23%,AUC为0.9837;黄斑水肿辅助诊断的灵敏度、特异性分别为99.02%、98.17%,AUC为0.9946;黄斑裂孔模型辅助诊断的灵敏度、特异性分别为98.91%、99.91%,AUC为0.9962;CNV模型辅助诊断的灵敏度、特异性分别为97.54%、94.71%,AUC为0.9875;AMD模型辅助诊断的灵敏度、特异性分别为95.12%、97.09%,AUC为0.9853。结论基于深度学习OCT图像眼底病变的眼底智能辅助诊断系统对于辅助诊断黄斑前膜、黄斑水肿、黄斑裂孔、CNV、AMD的诊断性能较高。Objective To establish an artificial intelligence robot-assisted diagnosis system for fundus diseases based on deep learning optical coherence tomography(OCT)and evaluate its application value.Methods Diagnostic test studies.From 2016 to 2019,25000 OCT images of 25000 patients treated at the Eye Center of the Second Affiliated Hospital of Zhejiang University School of Medicine were used as training sets and validation sets for the fundus intelligent assisted diagnosis system.Among them,macular epiretinal membrane(MERM),macular edema,macular hole,choroidal neovascularization(CNV),and age-related macular degeneration(AMD)were 5000 sheets each.The training set and the verification set are 18124 and 6876 sheets,respectively.Through the transfer learning Attention ResNet structure algorithm,the OCT image was characterized by lesion identification,the disease feature was extracted by a specific procedure,and the given image was distinguished from other types of disease according to the statistical characteristics of the target lesion.The model algorithms of MERM,macular edema,macular hole,CNV and AMD were initially formed,and the fundus intelligent auxiliary diagnosis system of five models was established.The performance of each model-assisted diagnosis in the fundus intelligent auxiliary diagnostic system was evaluated by applying the subject working characteristic curve,area under the curve(AUC),sensitivity,and specificity.Results With the intelligent auxiliary diagnosis system,the diagnostic sensitivity of the MERM was 93.5%,the specificity was 99.23%,and AUC was 0.9837;the diagnostic sensitivity of macular edema was 99.02%,the specificity was 98.17%,and AUC was 0.9946;the diagnostic sensitivity of macular hole was 98.91%,the specificity was 99.91%,AUC was 0.9962;the diagnostic sensitivity of CNV was 97.54%,the specificity was 94.71%,AUC was 0.9875;the diagnostic sensitivity of AMD was 95.12%,the specificity was 97.09%,AUC was 0.9853.Conclusions The artificial intelligence robot-assisted diagnosis system for fundus

关 键 词：人工智能 体层摄影术 光学相干 黄斑水肿 视网膜穿孔 脉络膜新生血管化 黄斑变性 黄斑前膜 

分 类 号：R77[医药卫生—眼科]