AbstractPurposeRetinal vessel oximetry is a promising tool to gain information on the microcirculation of the retina. Due to the ratio of reflectance’s of the vessel plane and the neighbouring tissue, computation of oxygen saturation values (SO2) varies between individuals. The aim of this study was to develop a data transformation approach to overcome pigmentation‐ dependency, allowing for inter‐subject comparison of SO2 values.MethodsArterial and venous SO2 results of 67 healthy subjects (mixed ethnicities) were related to the pigmentation value (RP) provided by the Vesselmap software (RVA Imedos, Jena). After normalization of the axis values, a regression line was rotated around the intercept with the y‐axis applying a rotation matrix (clockwise rotation).ResultsRegression lines before rotation were aSO2=0.808*RP+84.61 (R2 = 0.74) and vSO2 = 0.358*RP+39.17 (R2 = 0.56) for arteries and veins, respectively. Transformed arterial and venous regression lines were aSO2 = 0.066*RP+84.57 (R2 = 0.05) and vSO2 = 0.028*RP+39.09 (R2 = 0.01), respectively.ConclusionClockwise rotation of the regression lines for arteries and veins was able to reduce R2 from 0.74 to 0.05 for arteries and from 0.56 to 0.01 for veins, confirming that the data transformation resulted in retinal pigmentation independent SO2 results. The proposed approach might improve comparability of inter‐subject results and allow for determination of normative data.
Read full abstract