Abstract
Purpose :
Diabetic retinopathy causes capillary closure, which can be visualized in OCTA images. It may be useful to quantify the capillary closure apparent in these images. The purpopse of this study is to evaluate two different algorithms to measure the microvascular density in normal eyes and eyes with diabetic retinopathy (DR) imaged using ZEISS Cirrus AngioPlex.
Methods :
This was a prospective imaging study using a prototype CIRRUS 5000 AngioPlexy system (www.clinicaltrials.gov NCT02391585). Subjects with DR were imaged with OCTA and fundus photography. ETDRS grade was determined from the fundus image, and HbA1c and duration of diabetes were recorded. Normal eyes were recruited from AIBILI staff and patient companions. Both eyes of each subject were imaged using a 3x3 mm OCTA scan. Superficial Retinal Layer images were exported as bitmaps and processed with two algorithms to generate density maps, one based on the area of vessels as observed (area density), and one based on a map with vessels of one pixel width (length density). Vessel density was averaged over the full 3mmx3mm area. A separate region growing algorithm was used to determine the size of the foveal avascular zone (FAZ). A comparison to basic normal limits was done to determine if parameters could be used to separate normal eyes from DR.
Results :
Density and FAZ results from 47 eyes from 27 subjects with mean age of 65 (SD 8) with DR were imaged and 32 healthy eyes from 17 subjects with mean age of 50 (SD 14) are shown in Table 1. Also shown is the sensitivity to DR if the 95th percentile of the normal eyes is used as a threshold, which suggests length density is best at differentiating normal eyes from DR eyes. There is a range of density observed within each ETDRS grade level, suggesting that capillary closure may provide relevant information regarding progression in individual DR patients. This study did not detect a correlation between density or FAZ and HbA1c or duration of diabetes.
Conclusions :
OCT Angiography performed with Cirrus AngioPlex and using a length density algorithm provides a metric to monitor capillary closure in the parafoveal region.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.