Abstract
Purpose:
to explore the correlation between metrics of the cone mosaic imaged by an adaptive optics (AO) flood illumination retinal camera and retinal thickness in cross-section SD-OCT images in type 1 diabetes mellitus (DM1).
Methods:
AO retinal images of the photoreceptor mosaic were acquired in 16 DM1 patients; 16 healthy subjects were recruited as controls. Cone density, spacing of cones and packing arrangements of cones, estimated using Voronoi diagrams, were calculated within 1.5 degree eccentricity from the fovea along the vertical and horizontal meridians. From spacing of cones and Voronoi diagrams, two novel descriptors of the cone mosaic, the Linear Dispersion index (LDi) and the Heterogeneity Packing index (Pi), were calculated respectively. The subfoveal retinal thickness (SRT) was estimated in retinal cross-sections in SD-OCT images. Linear correlation analysis was performed to determine the relationships between cone metrics and SRT.
Results:
Patients had diagnosis of DM1 from 9 to 21 years earlier. Eight patients were diagnosed with no signs of diabetic retinopathy (noDR) and 8 with mild signs of DR (NPDR). The SRT was statistically significant thicker in DM1 eyes than controls (194±18 µm in NPDR, 199±31 µm in noDR and 153±36 in controls, p<0.05). Cone density showed no significant correlation with SRT (r=-0.30, p=0.09); the LDi approached statistical significance (r=0.35; p=0.05) and the HPi showed a high statistically significant correlation with SRT (r=-0.45, p=0.009).
Conclusions:
Cone density alone is not descriptive of the pathologic changes of the cones. The novel metrics based on spacing and packing arrangements of cones showed to correlate with the pathologic changes of foveal retinal thickness in patients with type 1 diabetes mellitus.