Chun Dal W, Bauer Robert M, Ward Thomas P, Dick John S B, Bower Kraig S
Ophthalmology Service, Walter Reed Army Medical Center, Washington, DC 20307, USA.
Mil Med. 2007 Apr;172(4):405-10. doi: 10.7205/milmed.172.4.405.
The goal was to evaluate a digital imaging system for diagnosing and grading diabetic retinopathy (DR) and cystoid macular edema (CME).
A single 45 degrees, nonmydriatic, digital color photograph was taken of 231 eyes of 120 patients with diabetes mellitus. The images were graded for DR and CME by a remote ophthalmologist, and the results were compared with dilated ophthalmoscopy performed by a retina specialist.
For DR, the level of agreement between digital image review and ophthalmoscopy was moderate (kappa = 0.44). The sensitivity and specificity of digital image review were 0.60 and 1.00, respectively. For CME, the level of agreement was moderate (kappa = 0.60). The sensitivity and specificity of digital image review were 0.60 and 0.99, respectively.
A single 45 degrees, nonmydriatic, digital image is not reliable as the sole modality for DR screening. However, with modifications, it may be useful where access to an experienced ophthalmologist is limited.
评估一种用于诊断和分级糖尿病视网膜病变(DR)及黄斑囊样水肿(CME)的数字成像系统。
对120例糖尿病患者的231只眼拍摄了一张45度的非散瞳数字彩色照片。由一名远程眼科医生对图像进行DR和CME分级,并将结果与视网膜专科医生进行的散瞳检眼镜检查结果进行比较。
对于DR,数字图像评估与检眼镜检查之间的一致性水平为中等(kappa = 0.44)。数字图像评估的敏感性和特异性分别为0.60和1.00。对于CME,一致性水平为中等(kappa = 0.60)。数字图像评估的敏感性和特异性分别为0.60和0.99。
一张45度的非散瞳数字图像作为DR筛查的唯一方式不可靠。然而,经过改进后,在经验丰富的眼科医生难以获取的情况下可能会有用。