Abstract
Purpose: :
The standard for evaluation of diabetic retinopathy is 7, 30°, 35mm stereo color slides based on the Early Treatment of Diabetic Retinopathy Study (ETDRS). Traditional retinal cameras use an external fixation target to manually guide field positioning. Some new systems incorporate an internal fixation target with preset field definitions. No study has compared variability between external and internal fixation target photography.
Methods: :
We compared 51 diabetic eyes consecutively photographed by the same certified photographer using a system with an external fixation target (Topcon TRC–50EX digital mydriatic camera) and a system with an internal fixation target (Topcon TRC–NW6S digital non–mydriatic retinal camera). Photography with the external fixation system followed ETDRS’ 7–field definition. Photography with the internal fixation system followed the IMAGEnet 2000 9–field definition. Mosaic images were then created from the fields. A reference mosaic for each field definition was also created. We measured retinal coverage of each study and reference mosaic to evaluate: 1) variability of coverage, 2) coverage compared to the reference mosaic coverage and, 3) agreement to reference field definitions.
Results: :
Mosaic photographs using an internal fixation target had less variable coverage compared to external fixation targets (internal fixation: 2.39 x106 pixels ± 0.06, 95%CI = 2.27, 2.50; external fixation: 2.01 x106 pixels ± 0.13, 95% CI = 1.76, 2.27). More study mosaics with an internal fixation target had the same coverage as the reference mosaic than external fixation mosaics (internal fixation: 30/51 eyes; external fixation: 1/51 eye). Among study mosaics with smaller coverage than its reference mosaic, internal fixation had a smaller coverage difference than external fixation (internal fixation: 90.69% to 99.8% of reference mosaic coverage; external fixation: 70.95% to 95.30% of reference mosaic coverage). Internal fixation mosaic photographs had a higher percentage of agreement with field definitions than external fixation mosaics (internal fixation: 97.26%, 95% CI = 93.19, 101; external fixation: 80.79%, 95% CI = 70.69, 90.89).
Conclusions: :
Camera systems using internal fixation targets provide more consistent retinal photography with less variable coverage and more retinal areas in agreement with field definitions. These are potential advantages in the application of computer algorithms to evaluate image quality, detect and quantify diabetic retinopathy pathology in telemedicine and clinical research.
Keywords: diabetes • imaging/image analysis: clinical • image processing