Purchase this article with an account.
Glen Y. Ozawa, Taras Litvin, Jorge A. Cuadros, Matthew S. Muller, Ann E. Elsner, Thomas J. Gast, Benno L. Petrig, George H. Bresnick; Comparison Of Flood Illumination To Line Scanning Laser Ophthalmoscope Images For Low Cost Diabetic Retinopathy Screening. Invest. Ophthalmol. Vis. Sci. 2011;52(14):1041.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To compare diabetic lesions in retinal images captured by Canon’s non-mydriatic retinal camera (CR-DGi) to Aeon Imaging’s Laser Scanning Digital Camera (LSDC).
Retinal images of 158 volunteer adult patients aged 54±10 were imaged with both the CR-DGi and LSDC at Eastmont Wellness Center, a diabetic retinopathy screening site within the EyePACS telemedicine network. All patients had been diagnosed with diabetes mellitus, with either known or suspected eye disease based on prior eye or primary care examinations. Each device captured three overlapping areas in the posterior pole of each eye. 22.8% of patients were dilated, due to a prior eye exam, or because the pupil was too small to obtain useable images with the CR-DGi. The retinal images were reviewed by EyePACS-certified graders.Images from the two cameras were compared side by side for cases in which diabetic retinopathy lesions were clearly visible. In addition, image artifacts caused by small pupils and unclear media in the vitreous or lens were compared. While the CR-DGi used flood illumination to obtain one image, the LSDC acquired confocal near-infrared images at 15.3 frames per second with a smaller minimum pupil (3 mm vs 3.7 mm in the CR-DGi small pupil mode).
Patients enrolled in the study included Caucasian Hispanics (47%), African Americans (26%), Asians (15%), Caucasian Non-Hispanics (9%) and others (3%). 46% of enrolled patients were female. Based on EyePACS grading of the Canon images, patients were diagnosed with no DR (7.0%); mild NPDR (29.9%); moderate NPDR (35.0%); severe NPDR (13.4%); and PDR (14.6%). 33.8% of patients were diagnosed with CSME in either eye.Diabetic lesions were visible with both devices. The LSDC’s illumination with longer wavelengths provided greater visibility of the deeper retina and choroid, and at times, masked the superficial lesions shown by the CR-DGi. However, the LSDC produced equal or superior images compared to the CR-DGi for patients with small pupils and media opacities. Retinal edema and optic nerve cupping were also more discernable with the LSDC.
Early field testing of the LSDC has demonstrated many cases of diabetic retinopathy lesions matching with the CR-DGi fundus camera. Interpretation of LSDC images will require grader retraining due to differences in the appearance of some lesions and the visibility of deeper retinal layers compared to traditional fundus photography.
This PDF is available to Subscribers Only