Adaptive optics imaging of cone photoreceptors and RPE cells based on detection of multiply scattered light from a wavelength-swept 3 MHz OCT light source

Abstract

Purpose : Multiply scattered light carries unique and independent contrast for assessing retinal cells such as cone photoreceptors, but is not typically collected in optical coherence tomography (OCT). The goal of this study is to show that multiply scattered light originating from the OCT light source itself can be recovered alongside conventional adaptive optics OCT imaging.

Methods : A custom-built multimodal adaptive optics imager based on a 790 nm scanning light ophthalmoscope (SLO) and a 3 MHz 1060 nm Fourier domain mode locked (FDML) OCT (AOSLO and AOOCT, respectively) was modified to collect multiply scattered light from both the 790 nm SLO and 1060 nm OCT light sources in a quadrant detection scheme (PMID33680539) using a multimode fiber bundle, while permitting confocal light to be collected for AOSLO/OCT. Nonconfocal split detection and darkfield images at 1060 nm were acquired and collected alongside AOSLO images at 790 nm using an analogous fiber-based quadrant detection scheme for comparison.

Results : Split detection and darkfield images of cone photoreceptor inner segments and retinal pigment epithelial (RPE) cells were successfully obtained using the 1060 nm FDML light source and were consistent with AOSLO images acquired at 790 nm across images from 4 healthy eyes. Nonconfocal split detection images of cone photoreceptors were qualitatively similar between 1060 and 790 nm. While foveal RPE cell mosaics could be delineated in darkfield images acquired at both 1060 and 790 nm, overall, the RPE cells were more readily visible when imaging using darkfield at 1060 nm compared to 790 nm. In these eyes, RPE cells could also be identified in 1060 nm darkfield images at eccentric locations out to 3 mm (cell spacing ranging from 12.7 to 14.3 µm).

Conclusions : While multiply scattered light is not currently compatible with OCT image formation, we demonstrate the possibility of using a quadrant detection scheme to collect this light simultaneously alongside conventionally acquired AOOCT volumes for multimodal imaging using the native OCT light source. Using a single FDML OCT light source to generate contrast that is typically only available in AOSLO systems may also allow for deeper imaging based on the longer wavelength.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.