Purchase this article with an account.
F. Orihuela–Espina, E. Claridge, I.B. Styles; Validation of a physics based model of the reflectance of the ocular fundus . Invest. Ophthalmol. Vis. Sci. 2004;45(13):2789.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: The long term goal is a novel imaging method which produces histologically informative images showing the spatial distribution and quantities of blood, Choroidal melanin, RPE melanin and Xanthophyll in the retinal fundus. The crucial requirement of the method is the ability to predict all instances of fundus colouration through physics based modelling. This paper describes the validation of the model against experimental data to establish whether measured spectra can be predicted by the model. The plausibility of the results is discussed. Methods: Ten experimental spectra of 4 Caucasian subjects, collected in the visible range by two different methods (Zagers et al. Appl Opt 41, 2002; Schweitzer et al. Oph Res 28, 1996) were provided by the above authors. The spectra include both foveal and perifoveal areas with and without venous sites. In the model, the structure of the fundus is represented as 3 layers comprising the following pigments: Xantophyll (receptor layer), melanin (Choroid and RPE) and haemoglobins (RPE). The properties of the layers are defined by specifying their thickness, refraction index, and the absorption and scatter coefficients and scattering phase functions of their components. The spectrum corresponding to a given instance of fundus tissue is computed using a Monte Carlo model of light transport from parameters defining pigment concentrations in each layer. The spectra cover the entire histologically plausible ranges for all four parameters, suitably discretised. For a given measured spectrum, the most closely matching model spectrum is found by searching a database of spectra pre–computed using Monte Carlo simulations. The selection criterion is to minimise the error, taken to be the variance of the differences between the log–s of the two spectra. Values of the best parameter set are taken to characterise the fundus tissue at the point of measurement. Results: A good fit was obtained for all the spectra (mean error 0.003, st. dev. 0.004). Mean concentrations (mmol/l) of Xanthophyll, RPE melanin, choroidal melanin and blood were 0.3, 5.44, 1.21, 6.64 in the fovea and 0.0, 5.23, 1.04, 5.91 in perifovea. Conclusions: The model of fundus colouration appears to correctly predict the spectral reflectance of the ocular fundus in foveal and perifoveal areas. Importantly, the model parameters returned by the validation are within the known physiological ranges and reflect the known physiological variations in the fundus tissue. The model is thus capable of quantitative analysis of fundus composition from the spectra. Work is in progress to compute the parameters from images, rather than from the spectra.
This PDF is available to Subscribers Only