Longitudinal repeatability of imaging perfused retinal capillaries in normal human and non-human primate eyes using adaptive optics

Abstract

Purpose : The ability to consistently image retinal vasculature in vivo has applications in the study of optic nerve head (ONH) and retinal disease. We assessed the intersession repeatability of imaging and quantifying perfused retinal capillaries in vivo in normal eyes using an adaptive optics scanning laser ophthalmoscope (AOSLO).

Methods : Confocal and split detector AOSLO reflectance videos of the retinal nerve fiber layer and superficial retinal vasculature were simultaneously acquired at the same depth in 2 normal rhesus monkeys and 3 normal human subjects in 2 or more sessions separated by ≥ 1 week. Locations ~100 to 850 µm from the ONH rim were imaged radially from the superotemporal to inferotemporal region. Perfusion images were created by computing the standard error of each pixel’s intensity over time within stabilized split detector videos. Major retinal vasculature was removed and remaining capillaries were traced. Resulting binary images of the same eye from different time points were registered using a cross-correlation technique. Capillary density (percentage of pixels with traced capillaries), mean capillary tortuosity (ratio of the actual length of an unbranched capillary segment to the linear distance between its endpoints) and mean length of unbranched capillary segments were computed in areas commonly imaged across sessions.

Results : Images of perfused vasculature were subjectively similar across sessions. Capillary density ranged from 2.0 to 3.4% in monkeys and 1.9 to 3.8% in humans with comparable Coefficients of Variation [CVs] (23.8% and 20.7%, respectively). Mean capillary tortuosity ranged from 1.323 to 2.686 (CV=26.8%) in monkeys and 1.177 to 1.801 (CV=15.2%) in humans. The mean length of unbranched capillary segments ranged from 61.8 to 114.6 μm (CV=20.1%) in monkeys and 111.7 to 167.7 μm (CV=14.8%) in humans. No significant differences in any parameter were found between sessions across monkey and human eyes (P>.05; Mann-Whitney).

Conclusions : Superficial perfused retinal capillaries can be consistently and noninvasively imaged in normal monkey and human eyes using an AOSLO. Mean capillary tortuosity and mean length of unbranched capillary segments provide complimentary information on capillary architecture (in addition to density) and can be applied to images acquired with other modalities (e.g. optical coherence tomography angiography).

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.