Oxygen Saturation Profiles in Retinal Vessels

Abstract

 

Purpose

 

To further our understanding of oxygen transport and metabolism in the retina, we use a quantitative model of oxygen transport and develop a new method to analyze retinal oximetry data. Using this approach we hope to ultimately increase the accuracy and spatial resolution of retinal oximetry measurements. This could improve our ability to measure oxygen saturation in vessels too small for analysis using current methods, which would be helpful when studying oxygen metabolism of the macula.

 

Methods

 

We designed and implemented a novel method to analyze raw image data from the Oxymap T1, a dual wavelength spectrophotometric retinal oximeter. By means of a custom-developed plugin for the image processing software ImageJ, our method produces spatially detailed maps of oxygen saturation in retinal vessels (Figure 1). To interpret the patterns observed, we applied finite element analysis and developed a computational model of oxygen transport in blood vessels using open source software (Elmer, gmsh, Paraview). The method was tested on data obtained from healthy volunteers.

 

Results

 

Retinal arterioles exhibit intravascular oxygen saturation profiles with maximum saturation in the center and lower saturation near the vessel wall, reflecting loss of oxygen by diffusion. For retinal venules the opposite is observed, with minimum saturation in the center, consistent with uptake of oxygen from tributaries or from tissue by diffusion. Arteriolar saturation profiles are generally symmetric, whereas major retinal venules mostly have asymmetric profiles with maximum saturation near the vessel wall facing the macula (Figure 2). The difference in saturation between the central/macular and peripheral halves of the superotemporal venule in thirteen healthy eyes was 7% saturation (mean, p<0.01). Fitting with these observations, our computational model predicts asymmetric saturation profiles downstream from a confluence of venules with differing oxygen saturation.

 

Conclusions

 

We have developed a new method to view and analyze detailed maps of retinal vessel oxygen saturation, as well as a computational model to interpret the patterns observed. We expect the method to contribute to our understanding of retinal oxygen metabolism in health and disease.

 

 

Pseudocolor map of oxygen saturation, superotemporal arcades.

 

 

Saturation vs. axial distance, principal venule from Fig. 1. Right half is macular side of vessel.

 

Keywords: 688 retina • 635 oxygen • 473 computational modeling