Application of deep learning image assessment software VeriSee™ for diabetic retinopathy screening
Application of deep learning image assessment software VeriSee™ for diabetic retinopathy screening
复制标题
DOI：
10.1016/j.jfma.2020.03.024
复制
发表时间：
2021-01-01
影响因子：
3.2
通讯作者：
Chen, Mingke
中科院分区：
文献类型：
--
作者：
Hsieh, Yi-Ting;Chuang, Lee-Ming;Chen, Mingke
目的：开发深度学习图像评估软件VeriSee（TM）并验证其对糖尿病视网膜病变（DR）严重程度分级的准确性。方法：回顾性招募2007年7月至2017年6月期间在国立台湾大学医院接受单视野、非散瞳、45度彩色视网膜眼底摄影的糖尿病患者。共收集了7524张可判断的彩色眼底图像，并由眼科医生对DR的严重程度进行分级。在这些图片中，将来自EyePACS数据集的5649张沿着以及另外31，612张彩色眼底图像用于VeriSee（TM）的模型训练。其他1875张图像用于验证，并由VeriSee（TM）、眼科医生和内科医生对DR的严重程度进行分级。结果：VeriSee（TM）诊断任何DR、可诊断DR和增殖性糖尿病视网膜病变（PDR）的AUC分别为0.955、0.955和0.984。VeriSee（TM）诊断DR和PDR的敏感性（分别为92.2%和90.9%）优于内科医生（分别为64.3%和20.6%）（两者均P < 0.001）。VeriSee（TM）对任何DR和可诊断DR的敏感性（分别为92.2%和89.2%）高于眼科医生（分别为86.9%和71.1%）（P < 0.001），而眼科医生的特异性更高。它可以提供临床援助，非眼科医生在DR筛查与非散瞳眼底照相。版权所有（C）2020，台湾医学会.由爱思唯尔台湾有限公司出版。
Purpose: To develop a deep learning image assessment software VeriSee (TM) and to validate its accuracy in grading the severity of diabetic retinopathy (DR).Methods: Diabetic patients who underwent single-field, nonmydriatic, 45-degree color retinal fundus photography at National Taiwan University Hospital between July 2007 and June 2017 were retrospectively recruited. A total of 7524 judgeable color fundus images were collected and were graded for the severity of DR by ophthalmologists. Among these pictures, 5649 along with another 31,612 color fundus images from the EyePACS dataset were used for model training of VeriSee (TM). The other 1875 images were used for validation and were graded for the severity of DR by VeriSee (TM), ophthalmologists, and internal physicians. Area under the receiver operating characteristic curve (AUC) for VeriSee (TM), and the sensitivities and specificities for VeriSee (TM), ophthalmologists, and internal physicians in diagnosing DR were calculated.Results: The AUCs for VeriSee (TM) in diagnosing any DR, referable DR and proliferative diabetic retinopathy (PDR) were 0.955, 0.955 and 0.984, respectively. VeriSee (TM) had better sensitivities in diagnosing any DR and PDR (92.2% and 90.9%, respectively) than internal physicians (64.3% and 20.6%, respectively) (P < 0.001 for both). VeriSee (TM) also had better sensitivities in diagnosing any DR and referable DR (92.2% and 89.2%, respectively) than ophthalmologists (86.9% and 71.1%, respectively) (P < 0.001 for both), while ophthalmologists had better specificities.Conclusion: VeriSee (TM) had good sensitivity and specificity in grading the severity of DR from color fundus images. It may offer clinical assistance to non-ophthalmologists in DR screening with nonmydriatic retinal fundus photography. Copyright (C) 2020, Formosan Medical Association. Published by Elsevier Taiwan LLC.