眼科 ›› 2021, Vol. 30 ›› Issue (5): 355-359.doi: 10.13281/j.cnki.issn.1004-4469.2021.05.006

• 论著 • 上一篇    下一篇

对糖尿病视网膜病变的三种人工智能诊断模型与人工检测的一致性分析

赵琦  杨文利  魏文斌  张永鹏  李蕾  张梦雨  郭婧   

  1. 首都医科大学附属北京同仁医院 北京同仁眼科中心 眼内肿瘤诊治研究北京市重点实验室 眼科学与视觉科学北京市重点实验室 医学人工智能研究与验证工信部重点实验室 100730

  • 收稿日期:2021-06-16 出版日期:2021-09-25 发布日期:2021-09-24
  • 通讯作者: 魏文斌,Email:weiwenbintr@163.com
  • 基金资助:
    首都卫生发展科研专项(首发2020-1-2052);北京市科委科技计划项目(Z201100005520045,Z18110000181 8003);北京市医院管理局“登峰”人才培养计划(DFL20150201)

Evaluation of the diagnostic accuracy in three kinds of artificial intelligence diagnosis model for detection of diabetic retinopathy on fundus photographs

Zhao Qi, Yang Wenli, Wei Wenbin, Zhang Yongpeng, Li Lei, Zhang Mengyu, Guo Jing   

  1. Beijing Tongren Eye Center, Beijing Key Laboratory of Intraocular Tumor Diagnosis and Treatment, Beijing Key Laboratory of Ophthalmology and Visual Sciences, Medical Artificial Intelligence Research and Verification Key Laboratory of the Ministry of Industry and Information Technology, Beijing Tongren Hospital, Capital Medical University, Beijing 100730, China

  • Received:2021-06-16 Online:2021-09-25 Published:2021-09-24
  • Contact: Wei Wenbin, Email: weiwenbintr@163.com
  • Supported by:
    The Capital Health Research and Development of Special (2020-1-2052); Science & Technology Project of Beijing Municipal Science & Technology Commission (Z201100005520045, Z181100001818003); the Beijing Municipal Administration of Hospitals’ Ascent Plan (DFL20150201)

摘要: 目的 比较三种基于卷积神经网络(CNN)的人工智能(AI)糖尿病视网膜病变(DR)眼底图像诊断模型与眼科医生诊断的一致性。设计 诊断试验。研究对象 北京同仁医院眼科236例(468眼)糖尿病患者的彩色眼底图像。方法  患者每眼分别以黄斑和视盘为中心免散瞳拍摄两张45°眼底图像。以眼科医生诊断结果作为金标准,对三种不同公司设计的AI DR诊断模型(模型1、2、3)进行DR有无检测、DR转诊检测、DR分期检测评价。主要指标 灵敏度(Se)、特异度(Sp)、受试者工作特征曲线(ROC)曲线下面积(AUC)。结果 DR有无检测:模型1灵敏度为 95.8%,特异度为90.1%,AUC为0.930;模型2灵敏度为96.5%,特异度为85.2%,AUC为0.908;模型3灵敏度为 96.2%,特异度为83.5%,AUC为0.917。DR转诊检测:模型1灵敏度为93.9%,特异度为90.1%,AUC为0.933;模型2灵敏度为97.7%,特异度为89.3%,AUC为0.935;模型3灵敏度为 95.4%,特异度为89.8%,AUC为0.926。DR分期检测:模型1灵敏度为 72.9%~90.1%,特异度为93.9%~97.8%,AUC为0.854~0.930;模型2灵敏度为68.8%~92.1%,特异度为90.6%~98.2%,AUC为0.831~0.914;模型3灵敏度为75.0%~83.5%,特异度为89.2%~96.8%,AUC为0.849~0.917。结论 三种AI诊断模型对糖尿病患者眼底图像DR有无及DR转诊检测的准确性均较高,可作为糖尿病患者DR筛查的辅助方法。(眼科, 2021, 30: 355-359)


关键词: 糖尿病视网膜病变/诊断, 眼底图像, 人工智能, 一致性

Abstract: Objective To compare the consistency of artificial intelligence (AI) diagnosis models and ophthalmologist in grading diabetic retinopathy (DR). Design Diagnosis tests. Participants 236 cases (468 eyes) of diabetic patients in Beijing Tongren Hospital were selected. Methods Two-field non-mydriatic macula- or optic disc-centered fundus photographs were performed. Using ophthalmologist diagnosis as the golden standard, three kinds of AI diagnosis model(model 1, 2, 3) based on Convolutional Neural Network (CNN) were evaluated, including DR yes/no, referable DR and DR stagingtests. Main Outcome Measures Sensitivity (Se), specificity (Sp), area under curve (AUC). Results The Se, Sp and AUC of DR yes/no were 95.8%, 90.1% and 0.930 on model 1, 96.5%, 85.2% and 0.908 on model 2, and 96.2%,83.5% and 0.917 on model 3, respectively. The Se, Sp and AUC of referable DR were 93.9%, 90.1% and 0.933 on model 1, 97.7%,89.3% and 0.935 on model 2, and 95.4%, 89.8% and 0.926 on model 3, respectively. The Se, Sp and AUC of stages of DR were 72.9%~90.1%, 93.9%~97.8% and 0.854~0.930 on model 1, 68.8%~92.1%, 90.6%~98.2% and 0.831~0.914 on model 2, and 75.0%~83.5%, 89.2%~96.8% and 0.849~0.917 on model 3, respectively. Conclusion Three kinds of AI diagnosis model have good sensitivity and specificity for the determination of DR and referable DR, and it could be used as an auxiliary tool for DR screening of diabetic patients. (Ophthalmol CHN, 2021, 30: 355-359)


Key words: diabetic retinopathy/diagnosis, fundus photography, artificial intelligence, consistency