Measurement of Optical Parameters of Turbid Media
Visible light and short-wave infrared (SWIR) light between 400 and 1600nm can penetrate into turbid samples, including most human soft tissues such as skin, with larger depth for longer wavelength and therefore offers potential spectral windows for functional imaging and medical monitoring without ionizing radiation hazards. A fundamental challenge in achieving medical application of the visible and SWIR light is to understand the relation between the optical response of the skin and its characteristic parameters defined with accurate optical models. The light propagation in a strongly turbid sample such as the skin can be accurately analyzed at the macroscopic scales (>~0.1mm) by the radiation transfer model originated from study of light transportation in atmosphere.
Our experimental study of light interaction with turbid media has been focused on the determination of four optical parameters, scattering coefficient ms, absorption coefficient ma, anisotropy factor g (assuming a HG function as the scattering phase function) and refractive index based on the radiative transfer and effective medium theories. To determine these optical parameters, one has to solve an inverse problem which requires accurate measurement of optical signals from a turbid sample and accurate numerical modeling. Our general approach is to use either an integrating sphere based system or a reflectance imaging based system to measure the distribution of scattered light signals from the sample excited with a monochromatic incident light beam followed by a Monte Carlo based modeling method to accurate simulate the light signals. Using the squared difference between the measured and simulated signals as an objective function, an iteration process is pursued by an inverse algorithm to modify the optical parameter values used in the Monte Carlo simulations until the objective function is minimized to satisfactory.
An automated reflectometer has been developed to measure the coherent reflectance Rc as a function of incident angle q. By comparing the measured Rc(q) curve with the calculated one based on the Fresnel equations for s- and p-polarized incident light beam, we can determine the complex refractive index n of a turbid sample.
The integrating sphere based method can be used as an in vitro method to acquire light signals such as diffuse transmittance Td, diffuse reflectance Rd and collimated transmittance Tc. When combined with the Monte Carlo modeling, this method can lead to a quick convergence in the inverse calculation to optimized values of optical parameters. Its disadvantage lies in the need of transmittance signals which often requires the samples be cut off and sliced into thin slabs. We have developed another reflectance imaging based method using one imaging detector such as a CCD camera. Compared to the integrating sphere method, it needs only one image of reflected light signals from the probed site and thus can be performed in a in vivo fashion without no requirement of sample preparation. However, the challenge to find an efficient inverse algorithm to guide the process of determination of optical parameters from one reflectance image data is quite challenging. We have demonstrated recently that the optical parameters can be uniquely determined from one reflectance image and search for more efficient inverse algorithm is underway.