Novel superpixel method to visualize fundus blood flow resistivity in healthy adults

Abstract

We aimed to perform superpixel segmentation of ocular blood flow maps obtained using laser speckle flowgraphy (LSFG) and investigate the effects of systemic parameters such as body weight, height, and sex on ocular blood flow resistivity. We studied 757 healthy participants (583 men, 174 women). We calculated the average beat strength over mean blur rate (BOM) as a LSFG resistivity index, as a function of age and sex using ordinary regions of interest (ROI) centered on the optic nerve head (ONH), the retinal vessels region and tissue around the ONH, and the choroid (CHD). We compared the ROI and superpixel-based methods, which are segmented based on image processing, for calculating the BOM. The sex differences in the BOM for the ONH, retinal-vessels region and tissue region of the ONH and CHD were significant for individuals aged ≤ 50 years (P < 0.01) but not those > 50 years old (P > 0.05). The average BOMs calculated using the ROI and superpixel methods were strongly correlated in the ONH (coefficient = 0.87, R² = 0.8, P < 0.0001, n = 5465). In summary, a superpixel-segmented BOM map is suitable for two-dimensional visualization of ocular blood flow resistivity.

Figure 1

Example of superpixel-segmented BOM maps. Representative maps of average blood flow and resistivity (BOM) obtained using laser speckle flowgraphy and simple linear iterative clustering. Images show the same healthy subject at ages 40 and 48 years. The average blood flow maps show no discernable change (cf. panels B-1 and C-1), but the BOM maps show an age-dependent increase in retinal blood vessel resistivity and choroidal resistivity (cf. panels B-2 and C-2). (A-1) is the fundus image of the same eye. (A-2) is a pixel-wise BOM map. BOM, average beat strength over mean blur rate; y.o., years old.

Figure 2

Scatter plots of BOM, (A) Optic nerve head (BOM.ONH) and (B) choroid (BOM.CHD), calculated using the rubber band method and the superpixel method. These scatter plots do not include BOM.T or TCR. The blue line on each graph was fitted using univariate regression. There were good correlations between the parameters calculated using the rubber band method and superpixel method for both regions (ONH: R² = 0.8, CHD: R² = 0.81). (C-1) is the fundus image of the same eye. Examples of average blood flow maps and BOM maps that showed high correlation (C-1,C-2) between the rubber band method and the supperpixel method are shown in panels (C-1) and (C-2); maps with low correlation are shown in (D-1) and (D-2). The blue arrow in (C-2) indicates an artifact caused by a shadow of the eyelash present during measuring.

Figure 3

Example ROI, regions of BOM analysis using LSFG. (A) Color map of total measurement area. An elliptical rubber band enclosed the entire optic nerve head (ONH) and a rectangular rubber band enclosed a region in the choroid (CHD). The ellipse surrounds the area of the ONH. (B) Software was used to demarcate the retinal vessels using an automated definitive threshold throughout the ONH, within the ONH vessels (shown in white) and the ONH tissue (black).

Figure 4

Diagram showing BOM calculation using standard rubber band selection method, representative example using an elliptical rubber band enclosing the entire ONH is shown. Functions Y and P are defined in Eqs. (1) and (4), respectively. MBR, mean blur rate; MBR_avg, temporal average of the MBR; BS, beat strength.

Figure 5

Superpixel topography in an LSFG map before and after SLIC. (A) Initial superpixel placement. (B) Final superpixel segmentation. (C) Initial superpixel and search region.

Figure 6

Diagram showing method for averaging BOM using the superpixel method with an elliptical rubber band placed on the ONH. The segmented superpixel figure illustrates the method rather than actual automatically segmented superpixels.