Quad-Fusion Adaptive Optics Scanning Light Ophthalmoscopy: A Novel Multi-Directional Non-Confocal Imaging Technique for Enhancement of Microvascular and Cellular Structures

Abstract

Purpose : Non-confocal adaptive optics scanning light ophthalmoscope (AOSLO) imaging reveals retinal structures such as microvasculature, hyalocytes, and photoreceptor inner segments. Visualization of these structures is highly dependent on the spatial relationship between the non-confocal detectors and the structure’s orientations. We present a novel AOSLO image processing solution which merges quadrantic non-confocal detector channels to visualize multi-directional features of microvascular and cellular structures unavailable with conventional techniques.

Methods : 20 patients with retinopathies and 5 healthy controls were imaged using a custom AOSLO featuring a quadrantic non-confocal detection array (PMID: 33680539). 1°x1° regions of interest of vasculature, hyalocytes, or photoreceptors structures were imaged. Quad-fusion images were generated by merging the 4 split-detection images using an emboss edge detection filtering approach (PMIDs: 35414987 & 36531581). Emboss filter masks were created using MATLAB (MathWorks, Natick, MA, USA) and convolved with the split-detection images. Filtered images were merged to form quad-fusion images using pixel intensity projection. Cell Identification was performed using imageJ’s TrackMate plugin.

Results : Quad-fusion AOSLO imaging revealed dynamic in vivo details of blood vessel walls, erythrocytes, hyalocytes, and photoreceptor inner segments independent of their orientation. Cell identification on quad-fusion images measured erythrocyte lineal density, hyalocyte density and velocity, and photoreceptor density. Multi-directional feature enhancement provided single cell details of focal vascular occlusions in sickle cell retinopathy and diabetic retinopathy; hyalocyte cell bodies and dynamic processes against the complex vitreous scafold; homogeneous intensity profile of photoreceptor inner segments, enabling rapid, accurate cell identification.

Conclusions : We performed the first orientation-bias-free quad-fusion AOSLO imaging by merging the non-confocal images after utilizing emboss edge detection filtering. We are able to reliably visualize microvasculature, erythrocytes, hyalocytes, and photoreceptor inner segments regardless of their orientation, providing quick, accurate, and objective quantification of cell morphology and functional characteristics for future clinical studies

This abstract was presented at the 2023 ARVO Imaging in the Eye Conference, held in New Orleans, LA, April 21-22, 2023.