Monte-Carlo simulation and tissue-phantom model for validation of ocular oximetry: supplement

Ocular oximetry, in which blood oxygen saturation is evaluated in retinal tissues, is a promising technique for the prevention, diagnosis and management of many diseases and conditions. However, the development of new tools for evaluating oxygen saturation in the eye fundus has often been limited by the lack of reference tools or techniques for such measurements. In this study, we describe a two-step validation method. The impact of scattering, blood volume fraction and lens yellowing on the oximetry model is investigated using a tissue phantom, while a Monte Carlo model of the light propagation in the eye fundus is used to study the effect of the fundus layered-structure. With this method, we were able to assess the performance of an ocular oximetry technique in the presence of confounding factors and to quantify the impact of the choroidal circulation on the accuracy of the measurements. The presented strategy will be useful to anyone involved in studies based on the eye fundus diffuse reflectance.

Monte-Carlo simulation and tissue-phantom model for validation of ocular oximetry: supplemental document 1 1

. DERIVE OXYGEN SATURATION FROM OXYGEN PARTIAL PRESSURE
The ABL90 Flex Plus gas analyzer -the gold standard for assessing blood oxygen levels in samples -was used as a co-oximeter. However, it is not recommended for measurements of samples with low hemoglobin concentration, as is the case in this study. Fortunately, the partial pressure measurement function remains valid as long as the turbidity of the sample is not high. The relationship between blood oxygen saturation (sO 2 ) and the oxygen partial pressure in the blood (pO 2 ) has been extensively studied and is well documented in literature [1][2][3][4][5]. This relationship is defined by the hemoglobin-oxygen dissociation curve, which obeys a sigmoid-like function, that can be shifted left or right according to temperature, pH, the concentration of 2,3-diphosphoglycerate and the concentration of dyshemoglobins [4,5].
In our experiments, the temperature was maintained at 37 • C while the concentrations of dyshemoglobins, and 2,3-diphosphoglycerate were negligible and therefore were not considered. The equation by Severinghaus [3] was used to derive sO 2 from pO 2 and the pH: where X = pO 2 × 10 0.48(pH−7.4)+0.0325 .
We nevertheless wanted to verify the applicability of the method to our samples. To do this, we prepared 40 samples for liquid phantoms, taking care to keep the hemoglobin and intralipid (IL) concentrations at levels acceptable for the oximetry function of the gas analyzer. Of these samples, 20 contained 4.5 g/dl hemoglobin and 20 contained 2.25 g/dl hemoglobin. The IL concentration for all 40 samples was maintained at 0.25% w/v and oxygen saturation was varied from 50% to 100%. In this way, we had readings of pO 2 and sO 2 by the gas analyzer; which allowed us to compare the sO 2 derived from pO 2 to that directly measured by the gas analyzer. The Figure S1 shows this comparison. The sO 2 derived from the pO 2 are in agreement with those measured by the gas analyzer. We can therefore, in the absence of sO 2 measurements from the gas analyzer, approximate with good accuracy the sO 2 values from pO 2 measurements. Fig. S1. This figure illustrates the linear correlation between the oxygen saturation derived from oxygen partial pressure and those directly measured by the ABL90 Flex Plus gas analyzer. Blue points are data points and the black line is the line of perfect agreement. Linear regression: slope was 0.991, bias was 0.0792, correlation coefficient was 0.997, p-value < 0.0001 and standard error was 0.0119.

Fig. S2
. sO 2 determined by least square fit on blood samples whose oxygen saturation varies between 15% and 100%. Different spectral ranges were used, and it was found that the range (530 nm, 585 nm) is the one which gives the best results over the whole range of saturations.  . Variation in oxygen partial pressure between replicates of the same experiment. Oxygen saturation was measured on the red blood cells using the gas analyzer. The oxygen partial pressure was measured in the solution whose IL concentration is 0.25% w/v (Table S2)  g(−) 0.9 0.9 0.9 0.9 Table S1. Overview of optical parameters used in Monte-Carlo simulations. RPE and choroidal melanin concentrations were set to 400 mmol/l and 100 mmol/l respectively, the choroidal blood oxygen saturation was 95%, the BVF in the neural retina was 3% v/v and the retinal blood oxygen saturation was set to 55%.