Abstract
Purpose: :
To compare a novel strategy for the quantitation of drusen area in eyes with non-exudative age-related macular degeneration (AMD) using spectral domain optical coherence tomography (SDOCT) imaging compared with drusen area using color fundus photography.
Methods: :
23 eyes from 16 patients with drusen secondary to non-exudative AMD were enrolled in an ongoing prospective SDOCT study at the Bascom Palmer Eye Institute. These eyes were imaged using the Cirrus SDOCT (Carl Zeiss Meditec, Dublin, CA). SDOCT datasets from each eye consisted of 40000 uniformly spaced A-scans organized as 200 A-scans per B-scan in a 6mm X 6mm area comprised of 200 horizontal B-scans. The raster scan was centered on the fovea. Quantitative measurements of drusen area were obtained by applying a novel algorithm to the full dataset as well as data points contained within 3 mm and 5 mm circles centered on the fovea. Drusen are only identified using this algorithm if they elevate the RPE by at least 19 microns. These drusen areas were compared to drusen areas quantified by grading color fundus images after exporting the files to a CintiQ WACOM digitizing tablet (WACOM Corp., Vancouver, WA) and the drusen boundaries were drawn by hand.
Results: :
The mean drusen area obtained by manually outlining drusen on the color fundus images was significantly larger than the drusen area detected by the SDOCT algorithm for the 3mm circle (p=0.006) and the 5mm circle (p=0.007). The average difference between the drusen area detected on the color fundus images and the area detected by the SDOCT algorithm was 0.53mm2 for the 3mm circle and 1.07 mm2 for 5mm circle. The Bland-Altman limits of agreement (LOA) on the difference between the two methods ranged from -1.0 to +2.1 for the 3mm circle and -1.4 to +3.5 for the 5mm circle. The width of the 3mm LOA was 52% of the range of the 3mm measurements and 47% of the range of the 5mm measurements.
Conclusions: :
Significantly different drusen areas were obtained using color fundus images and SDOCT datasets. The SDOCT algorithm only detected drusen that significantly deformed the RPE geometry ,so the drusen areas obtained from color fundus images would be larger in most cases since these areas would include drusen regardless if they were flat or elevated. These two approaches offer complementary information with SDOCT offering the ability to reliably assess drusen area and volume for a subset of drusen that are elevated compared with the normal RPE contour. This drusen subset may be useful as a clinical trial endpoint when assessing therapies for the treatment of non-exudative AMD.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • age-related macular degeneration • imaging/image analysis: clinical