Comparison of cone mosaic metrics between the Spectralis high magnification module (HMM) and adaptive optics scanning light ophthalmoscopy (AOSLO)

Abstract

Purpose : Non-invasive imaging of the cone mosaic is typically achieved using adaptive optics (AO)-based devices. However, in eyes with good optical quality the cone mosaic can be resolved without AO. The Heidelberg Spectralis HMM can visualize ocular microstructures resembling cones. Here we compared cone mosaic metrics from AOSLO images to those from Spectralis HMM images.

Methods : Subjects with normal vision and previously acquired AOSLO images had Spectralis wide-field and HMM imaging performed on one eye (Fig. 1). At least 10 HMM images (8° field of view) were obtained at superior and temporal locations ranging 3.5-8.5° from the fovea. Individual HMM images were aligned using affine registration (i2k Retina, Dual Align, LLC). Between 2-20 HMM images were averaged using Image J. A semi-automated cell counting algorithm (Mosaic Analytics, Translational Imaging Innovations) was used to identify photoreceptors within 200 x 200 µm regions of interest at similar eccentricities from the fovea in the AOSLO and the averaged HMM images. Cell density and spacing were calculated from these coordinates and agreement was assessed statistically.

Results : Ten subjects (age range 24-34 years, 1 M, 9 F) had HMM imaging 0.5-6.7 years after their AOSLO images (mean 1.85 years). The visualization of cone mosaic was not uniform across the HMM images. All density and spacing metrics were significantly different between the AOSLO and HMM (Fig. 2). In addition, the level of disagreement increased as the cell density increased.

Conclusions : Though ocular microstructures resembling the cone mosaic can be visualized in vivo using the Spectralis HMM, there remain discrepancies in cone mosaic metrics. Nevertheless, there may be value in utilizing this system to screen subjects for follow-on AOSLO.

This is a 2020 Imaging in the Eye Conference abstract.

 

Figure 1. Differences in retinal images acquired with Spectralis HMM and AOSLO. Shown is a 30° FOV Spectralis near infrared reflectance fundus image (A) and an 8° FOV HMM image from the same subject. A 200 x 200 μm ROI outlined in (B) from the Spectralis HMM image is shown in (C). The corresponding split-detector AOSLO ROI from the same retinal location is shown in (D). Scale bar for panels C and D is 50µm.