Purchase this article with an account.
Marco Lombardo, Maria Cristina Parravano, Daniela Giannini, Sebastiano Serrao, Lucia Ziccardi, Giuseppe Lombardo; Novel adaptive optics imaging biomarkers to investigate the early pathologic changes of the cone mosaic in patients with type 1 diabetes mellitus. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):4923.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate the accuracy of adaptive optics (AO)-based imaging biomarkers to assess the spatial arrangement of cones in patients suffering from type 1 diabetes mellitus (DM1).
An AO flood-illumination retinal camera was used to obtain images of the cone mosaic in 16 DM1 patients, aged 22-57 years old. Eight DM1 patients had mild non proliferative diabetic retinopathy (NPDR) and eight patients had no sign of diabetic retinopathy (noDR); 16 healthy volunteers were recruited as controls. Cone density, spacing between cones, and the preferred packing arrangements of cones were assessed in 160x160 µm sampling areas at 1.5 degree eccentricity from the fovea along all retinal meridians. The cone spacing and preferred arrangements of cones were calculated using the Nearest Neighbours Distance (NND) and the Voronoi diagrams respectively. The novel metrics included the linear dispersion index (LDi) and the heterogeneity packing index (HPi). The LDi is the ratio between the standard deviation and the mean of the NND; the HPi represents the fraction of hexagonal Voronoi tiles over the non-hexagonal tiles and is expressed in percentage. Logistic regression analysis was performed to identify patients with DM1 using cone density, LDi and HPi parameters as descriptors.
The cone density was lower in DM1 patients than controls, though the differences reached statistically significance only between NPDR cases and controls (P<0.001). The LDi was higher in DM1 patients than controls, though the difference was statistically significance only between NPDR cases and controls (P=0.01). The HPi was statistically significantly lower in patients with DM1, both NPDR and noDR cases, than controls (P≤0.01). Logistic regression analysis achieved 94% classification accuracy for DM1 patients based on the complementary use of the AO-based imaging biomarkers.
The complementary use of cone metrics shows great potential to detect early pathologic changes of the cone mosaic in patients affected by diabetes mellitus.
This PDF is available to Subscribers Only