Abstract
Purpose :
To validate the performance of a fully automated choroidal segmentation algorithm and to analyze the relationship between the patterns of choroidal thickness and choroidal blood vessel distribution in eyes with dry AMD imaged using widefield swept-source OCT (SS-OCT).
Methods :
Patients were enrolled in a prospective OCT study at Bascom Palmer Eye Institute. Eyes with a diagnosis of dry age-related macular degeneration (AMD) were imaged using a prototype 100-kHz SS-OCT instrument (Carl Zeiss Meditec, Dublin, CA) with a central wavelength of 1,050 nm. We used an OCT cube scan pattern (512x512 A-scans) covering a 12x12mm retinal area. Autofluorescence (AF) images (Heidelberg Spectralis) were also acquired. The eyes were partitioned into 3 groups based on the assessment of (presence/absence/inconclusive) of reticular pseudodrusen (RPD) on both the AF images and the OCT images. RPD on the OCT images were assessed by examining en face slabs with boundaries from 35 to 55 µm above the RPE. Five eyes were randomly chosen from each of the three groups. The boundaries of the choroid were manually segmented in a total of 15 scans from 15 eyes and compared with the results of the automated algorithm.
Results :
Pointwise choroidal thickness maps were automatically generated and compared to manually generated choroidal thickness maps over the entire scan area. The average difference between the two maps was 13.7 microns. Although subfoveal choroidal thickness is typically thinner in eyes with RPD compared with eyes without RPD, we found the specific thickness patterns varied significantly from eye to eye. In particular, choroidal thickness appeared to correlate with the geographic distribution of the large choroid vessels.
Conclusions :
When applied to widefield SS-OCT scans, our fully automated choroidal segmentation algorithm was shown to be relatively close pointwise to the manually drawn boundaries and is quite able to capture the different patterns of choroidal thickness over a wide area.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.