Spectroscopic analysis of scattering media via different quantification techniques
Huong, Audrey Kah Ching (2012) Spectroscopic analysis of scattering media via different quantification techniques. PhD thesis, University of Nottingham.
This thesis outlines problems in the estimation of concentration value of an absorber present in a medium that is scattering and shortcomings of using the currently available techniques for the characterisation of such a scattering medium. The work describes experiment and analytic techniques used to overcome these problems. This thesis explored experimentally the practicality of using polarisation subtraction technique to minimise the effects of scattering components on the measured data. This work also considered the case when this technique has not been used owing to its limited application, and to that end, the improved linear equation and nonlinear fitting models, and gradient processing method were developed on the basis of the assumed behaviour of how a medium’s scattering coefficient changes with wavelength to provide information about the fractional concentration value of an absorber. The performance of these techniques evaluated via simulation showed that linear equation model has the fastest processing speed, nonlinear fitting method is robust to system noise and is able to provide an overall more accurate estimate of value with mean of errors of less than ±1%. The gradient processing method has intermediate performances with accuracy of its estimated value improved by about 30% with an increase in the spectral resolution from 1 nm to 0.5 nm. This work concludes that gradient processing method would be employed if accuracy of estimated value, noise robustness and computing time are of concern. However, nonlinear fitting method would be chosen in case high accuracy of the estimated value is required. Both of these methods can be suitably used as complementary techniques to clinical assessment of skin grafts and burnt skin. The simultaneous solution of linear equation model works well if all the measurement parameters are known.
Actions (Archive Staff Only)