Toward a clinical optoretinogram: a review of noninvasive, optical tests of retinal neural function

Abstract

The past few years have witnessed rapid development of the optoretinogram-a noninvasive, optical measurement of neural function in the retina, and especially the photoreceptors (Ph). While its recent development has been rapid, it represents the culmination of hundreds of experiments spanning decades. Early work showed measurable and reproducible changes in the optical properties of retinal explants and suspensions of Ph, and uncovered some of the biophysical and biochemical mechanisms underlying them. That work thus provided critical motivation for more recent work based on clinical imaging platforms, whose eventual goal is the improvement of ophthalmic care and streamlining the discovery of novel therapeutics. The first part of this review consists of a selective summary of the early work, and identifies four kinds of stimulus-evoked optical signals that have emerged from it: changes in light scattered from the membranous discs of the Ph's outer segment (OS), changes in light scattered by the front and back boundaries of the OS, rearrangement of scattering material in and near the OS, and changes in the OS length. In the past decade, all four of these signals have continued to be investigated using imaging systems already used in the clinic or intended for clinical and translational use. The second part of this review discusses these imaging modalities, their potential to detect and quantify the signals of interest, and other factors influencing their translational promise. Particular attention is paid to phase-sensitive optical coherence tomography (OCT) with adaptive optics (AO), a method in which both the amplitude and the phase of light reflected from individual Ph is monitored as visible stimuli are delivered to them. The record of the light's phase is decoded to reveal a reproducible pattern of deformation in the OS, while the amplitude reveals changes in scattering and structural rearrangements. The method has been demonstrated in a few labs and has been used to measure responses from both rods and cones. With the ability to detect responses to stimuli isomerizing less than 0.01% of photopigment, this technique may prove to be a quick, noninvasive, and objective way to measure subtle disease-related dysfunction at the cellular level, and thus to provide an entirely new and complementary biomarker for retinal disease and recovery.

Keywords: Optoretinography (ORG); adaptive optics (AO); functional imaging; optical coherence tomography (OCT); photoreceptors (Ph).

Figure 1

Overview of retinal anatomy, phototransduction, and retinal imaging using OCT and AO-OCT. (A) A cross-sectional diagram of the eye, as viewed from above. An image is formed on the retina when the eye’s two main optical elements—the cornea and lens—focus the incident light, while the iris controls the amount of light entering the eye. The center of the retina, called the fovea, is specialized for high acuity vision, and is thus densely packed with cone Ph. (B) Optical coherence tomography (OCT) is a standard of ophthalmic care, providing clinicians with a cross-sectional image of the retina. Shown here is a scan through the macular region, including the fovea and foveal pit. Clinical OCT systems have axial resolution on the order of 10 μm, sufficient for resolving the layers of the neural retina. Light incident on the retina must penetrate the inner neural layers into the outer retina, where the Ph lie. The Ph send signals to the BPC, which in turn signal the GC, whose axons form bundles in the NFL. These bundles meet to form the optic nerve, and carry the retinal signal to the brain. While the main neural layers are resolved in clinical OCT images, lateral resolution is insufficient to individuate cells. (C) A cross-sectional image acquired with AO-OCT. AO provides the lateral resolution required to observe single cells and subcellular features. This AO-OCT image shows two correlated arrays of scattering objects, believed to be the inner and outer boundaries of the photoreceptor OS. (D) These boundaries are called the IS/OS and COST or ROST. The OS consists of a cylindrical, membrane-bound stack of between 200 and 1,000 circular membranous discs, studded with light-sensitive proteins called opsins. The capture of photons by opsins in the OS is the first step in seeing. OCT, optical coherence tomography; AO, adaptive optics; Ph, photoreceptors; BPC, bipolar cells; GC, ganglion cells; NFL, nerve fiber layer; OS, outer segments; IS/OS, inner segment outer segment junction; COST, cone outer segment tips; ROST, rod outer segment tips.

Figure 2

Hypothetical origins of light-evoked changes in coherent flood illumination and SLO images of cones. Jonnal et al., 2007 reported the first observations of stimulus-evoked changes in the intensity of areal images of the cone Ph, using an adaptive optics flood illumination camera (79). They showed that the resulting oscillations in cone reflectance were dependent on the coherence of the illumination source, and hypothetically attributed them to elongation of the cone OS in the presence of coherent interference between light reflected by its two ends, the IS/OS and COST. The temporal coherence function [ε(z)] of a source describes how the phase of a light beam decorrelates as it propagates through space and, thus, how the amplitude of an interference fringe falls as the interfering scatterers are axially displaced. The coherence function is usually summarized by its FWHM. After Jonnal et al., 2007, several investigators (80-82) reported similar oscillations in cone reflectance, using sources with coherence lengths shorter than the OS length of the cones. Two hypothetical explanations for the are shown above. At the top are plots of coherence functions based on several reported bandwidths or coherence lengths (solid lines). At the bottom are three cartoon cone outer segments, showing the primary reflectors at IS/OS and COST, as well as additional OS reflections observed in many cones. The coherence functions plotted with solid lines are Gaussian functions with width parameters derived from the reported FWHM. FWHM is a convenient measure of source coherence because the source spectrum and coherence functions have a simple relationship when they are Gaussian. In practice, however, neither spectra nor coherence functions are Gaussian. The dotted and dashed lines show the coherence functions that would result if Rha et al.’s (80) spectrum were rectangular in shape or Gaussian with a small amount of noise, respectively, but with the same spectral FWHM. Departures from a Gaussian shape can have significant effects in the tails of the coherence function, and could lead to high fringe amplitude even when none is predicted. Complicating matters further, as reported by Azimipour et al., 2019, the photoreceptor OS often has additional reflections (depicted in red in the cartoons), observed to move and change amplitude upon stimulation. These additional reflections are likely to contribute to the en face measurement, significantly complicating quantification even when coherence is shorter than the OS. SLO, scanning light ophthalmoscopy; Ph, photoreceptors; OS, outer segments; IS/OS, inner segment outer segment junction; COST, cone outer segment tips; FWHM, full-width at half maximum.

Figure 3

Photoreceptor deformations evoked by light stimuli, measured with phase-sensitive OCT with AO or DAC. As has been shown with both full-field and scanning OCT systems with AO or digital aberration correction, the length the photoreceptor outer segment changes in response to light stimuli. These systems have spatial resolution sufficient to individuate cone (and sometimes rod) Ph. In the red inset is shown a small portion of an en face projection through the rods and cones, collected using our AO-OCT-SLO system. The green and blue boxes highlight single cone and rod Ph. The plot depicts the concepts and key features of the cone (green) and rod (blue) responses to flashes bleaching 12% and 0.8% of photopigment, respectively. The cone response (l_OS, green) exhibits an initial contractile phase (inset) followed by rapid elongation and slower return to baseline. The amplitude of the rod response (blue) is greater relative to bleaching, but rises and falls more slowly than in cones. An initial contractile phase has not yet been observed in rods. Several parameters are labeled on the cone curve: maximum contraction and elongation (l_min and l_max, respectively), as well as slopes Δl/Δt for the contraction, elongation, and recovery portions of the curve. All of these parameters appear to be dependent on the number of opsin photoisomerizations, in both rods and cones. The dose dependence of time-to-rise t_max has yet to be established; in some experiments it appears to be dose independent. Each parameter may have unique sensitivity and dynamic range, both of which are critical for assessing translational impact. These biomarkers should be studied carefully in healthy subjects before designing disease studies or endpoints. Conceptual plots based on unpublished data from separate experiments in my lab using (I) full-field AO-OCT to measure L-/M-cone responses to a flash bleaching 12% of each cell’s photopigment; (II) confocal, scanning AO-OCT to measure rod responses to a flash bleaching 0.8% of rhodopsin. OCT, optical coherence tomography; AO, adaptive optics; DAC, digital aberration correction; Ph, photoreceptors; SLO, scanning light ophthalmoscopy.