Adaptive optics imaging of putative cone inner segments within geographic atrophy lesions

Abstract

 

Purpose

 

The time course of cone loss relative to the formation of geographic atrophy (GA) lesions in age-related macular degeneration is not well characterized. Confocal adaptive optics scanning light ophthalmoscopy (AOSLO) images of GA lesions contain hyper-reflective cone-like features, but such images are difficult to interpret in advanced disease states such as GA. Non-confocal split-detector imaging in AOSLO has been shown to reveal structures consistent with cone inner segments (Scoles et al. 2014). Here we use two non-confocal AOSLO methods to determine whether putative cone inner segments exist within GA lesions and if they correlate with the cone-like structures seen in confocal AOSLO images.

 

Methods

 

Six patients were imaged, five with near-infrared (NIR) light in a compact commercial AOSLO prototype modified for split-detector imaging. The central portion of the PSF was directed to one detector for confocal imaging and a knife edge prism split the remaining light into two additional detectors. Five patients were imaged with visible light in a research AOSLO with a novel non-confocal method based on offset aperture (Chui et al. 2012) and split-detector methods. At each location, images were sequentially obtained with a ~10 Airy disk aperture offset by ~6 Airy disks to 8 equally spaced radial positions. Images were co-registered with simultaneously acquired NIR images. Registered images from each position were averaged and combined to enhance contrast. Identical locations were imaged in both instruments in some eyes permitting method comparison.

 

Results

 

Structures consistent in appearance to putative cone inner segments were observed within some small GA lesions. Cone-like features in confocal AOSLO images did not always co-localize with putative cone inner segments, suggesting that some hyper-reflective features in confocal AOSLO arise from light scattered by other structures. Conversely, some areas where cones were not visible in confocal AOSLO contained putative cone inner segment structures in non-confocal images.

 

Conclusions

 

Some cones survive in small GA lesions. Images similar to those obtained with the split-detector method can be obtained using a single detector. Cone survival in small GA lesions offers hope that future treatments aimed at restoring RPE health may rescue remaining cones. Further study is needed to understand the rate of cone survival after RPE cell loss is initiated in AMD.