Pilot investigation of multimodal and multiscale retinal imaging technology developed by a trans-European multidisciplinary consortium

Abstract

Purpose : Adaptive optics (AO) has enabled visualizing cellular structure in the retina, yet its clinical usability has been hampered by small field-of-view and bulkiness of instrumentation. We developed a compact imaging system with a multiscale optical architecture that delivers large views of the retina as well as highly-magnified AO-corrected images. This study explored imaging performance in human eyes.

Methods : The prototype included three imaging modalities: confocal scanning laser ophthalmoscopy (SLO) based on 786 nm super-luminescent diode scanned at 8 kHz, optical coherence tomography (OCT) based on a 1050 nm swept-source at 200 kHz, and OCT angiography (OCTA). Each modality could be operated with a field-of-view of either 40 x 30 deg or 4 x 3 deg. In the latter case, ocular wavefront aberrations were corrected by AO operating in closed loop at 13 Hz. During OCT/OCTA examinations, head and eye movements were compensated by automated pupil and retinal tracking at 100 Hz. The ability to visualize microscopic retinal structure was experimented in 35 healthy volunteers and 39 patients affected by various fundus pathologies, including diabetic retinopathy, age-related macular degeneration and inherited retinal dystrophies.

Results : The overall dimensions of the system were 49 x 50 x 50 cm. Large-field SLO and OCT provided quick overviews of the retina while both AOSLO and AOOCT enabled the visualization of parafoveal cone cells, nerve fiber bundles, capillaries, and microscopic lesions. Multiple-averaged AOOCT B-scans revealed Müller-cell-like structures. AOOCTA could image capillaries and choriocapillaris with minimal shadowing artifacts. In pathological cases, AOSLO showed the distribution and lateral extent of microscopic lesions like micro-aneurisms, micro-exudates, atrophic region borders, pigment clusters, and photoreceptor losses. AOOCT scans completed these observations with information on depth location and axial extent of these alterations.

Conclusions : The multimodal multiscale imaging architecture enabled in-depth retinal examinations through a straightforward process: 1) acquire an overview image, 2) identify regions of interest, 3) examine these regions with near-histology 3-D resolution. When confirmed by further investigations, this technology should offer new possibilities for retinal phenotyping and clinical research in new therapies.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.