|
Integrating spheres (IS) facilitate accurate measurements of the total reflectance and transmittance of turbid media, which can be used to determine optical properties of the sample measured. Translation of measurements into optical properties are achieved using theoretical photon migration models. A widely used approach with IS measurements is to use the inverse adding-doubling (IAD) method that utilizes the forward adding-doubling method, which is a rigorous numerical forward solver of the 1-D radiative transport equation. In order to experimentally satisfy the 1-D nature of the theoretical model, samples must be large enough to be modeled as infinite in extent along axes normal to incident beam. Here, we explore constraint on the required sample dimensions by comparing errors in modeled reflectance and transmittance between the adding-doubling and Monte Carlo simulations. We compare both the forward predictions and the inverse extraction of the optical properties for samples with varying dimensions, sample optical properties and beam profiles. Lateral losses (loss of light from sides of the sample) were observed to be significant when illumination beam diameters become comparable to sample length. Errors of 2-3% were noted between MC predictions vs. the adding-doubling estimates for reflectance and transmittance and these translated to 5-30% errors in IAD estimated optical absorption while the extracted scattering coefficients remained unaffected and had errors < 2%, relative to simulated values. We find that when the incident beam had diameter less than 80% of the sample length, the estimated optical properties of the medium were well extracted using the IAD.
|
