Purpose:
To demonstrate that current SD-OCT macular analysis algorithms may fail to detect the presence of subretinal drusenoid deposits (SDD) that are more readily detected with high-resolution line raster B-scans and/or curved en face reconstruction slabs (C-scan).
Methods:
An observational case series of 5 eyes comparing the ability of various commercially available SD-OCT scanning systems and protocols to detect the presence of SDD.
Results:
In these 5 eyes, SDD were missed with lower resolution line scans used in standard macular cube protocols. Consequently, the drusen maps generated from these cube scans failed to demonstrate the presence of SDD (A). Manual segmentation on a C-scan slab ("RPE" algorithm and manual segmentation thickness of 37 µm moved above the RPE) clearly detected SDD (B; see arrows pointing to representative structures). High resolution averaged line B-scans were the most sensitive in detecting the presence of SDD but were limited by acquisition time and typically wide B-scan spacing.
Conclusions:
The presence of SDD in eyes with age-related macular degeneration is considered a risk factor for conversion to advanced AMD. Conventional SD-OCT scanning protocols are not reliable in detecting the presence and distribution of these important lesions due to the segmentation algorithm optimization for sub-RPE structures (C; SDD = triangles, Druse = circle, RPE = black line, RPE algorithm = red line). En face curved C-scan slabs cutting above the RPE contour demonstrate the distribution of these lesions, but this technique is currently limited by the lack of automation and resolution of current commercially available SD-OCT systems. With the advent of future medications aimed at altering the natural history of non-exudative AMD, sensitive, automated, and quantitative tracking of its course will become paramount.
Keywords: imaging/image analysis: clinical • age-related macular degeneration • drusen