Abstract
Purpose :
Oxygenation of biological tissue plays a key role in maintaining physiological activity and life. Local oxygenation abnormalities occur at the onset or with the development of many pathologies. We have developed a technique for measuring oxygen saturation (StO2) at targeted locations in the eye fundus, based on the derivation of StO2 from broadband visible light diffuse reflectance spectroscopy. In this study, we assess the sensitivity of this technique to spectral acquisition parameters such as spectral band, spectral resolution, proportion of the acquisition area occupied by a blood vessel, illumination light intensity and noise.
Methods :
Monte-Carlo simulations of light propagation in a 4-layers model of the retina were performed. Each parameter evaluated was varied, while all others were held constant. Light intensity was simulated by varying the number of photons from 100 thousand to 6 million photons. The proportion of the acquisition area occupied by a blood vessel was simulated by varying the acquisition volume for a given vessel size. Spectral resolution was varied by convolving high resolution spectra with a Gaussian function of variable width. Gaussian noise of varying standard deviations was added to the spectra. Vessel StO2 was varied from 0% to 100%. Finally, calculated StO2 values were compared to the actual values used in the simulations.
Results :
The spectral band giving the best accuracy over the entire range of StO2, after correcting for the effects of light scattering, was 530 nm to 585 nm. StO2 accuracy dropped drastically when the spectral resolution was below 4 nm. An acquisition area larger than a blood vessel led to an underestimation of StO2 in the vessel, especially for high actual StO2 values. There was a threshold above which an increase in light did not have a significant impact on the variability of the calculated StO2. A linear correlation was found between the additive noise level and the variability in calculated StO2.
Conclusions :
The use of a broad spectral band allows an accurate assessment of StO2 in the retinal microcapillaries, opening the door to the study of local changes in hemodynamics. To minimize the effect of various confounding factors, it is essential to make an informed choice of acquisition parameters, many of which are brought to light in this study by the use of a tissue-light interaction model.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.