ESEL Paper Review_20110117 by Aamir Alaud-din
1. Title and Author
Title: Handling of first-order Rayleigh scatter in PARAFAC modeling of fluorescence excitation-emission data
Authors: Asmund Rinnan1, Charlotte M. Andersen*
1The Royal Veterinary and Agricultural University, Department of Dairy and Food Science, Food Technology, Rolighedsvej 30, DK-1958 Frederiksberg Denmark
*The Royal Veterinary and Agricultural University, Department of Dairy and Food Science, Food Technology, Rolighedsvej 30, DK-1958 Frederiksberg Denmark
Summary of Paper
Fluorescence, a type of spectroscopy, also got its application in food science, analytical chemistry, biochemistry, environmental science and engineering and others. In this paper, handling of first order Rayleigh scatter in PARAFAC modeling is discussed. For this purpose, three different data sets were used. First data set included 15 samples of varying compositions prepared with Catechol, Hydroquinone, Indole, Tryptophane and Tyrosine were used. Second data set consisted of 16 samples of varying compositions, prepared with Hydroquinone, Tryptophane, Phenylaniline and Dopa was used. The third sample was the sugar sample taken from the final unit operation from a sugar plant.
In data pretreatment process, first data set was reduced by removing all the emission wavelengths which were greater than 440nm. Excitations from 200 to 230nm were removed in order to reduce the amount of noise. All the emission wavelengths greater than 440nm were removed from the third data set. The data in the EEM where the emission was less than excitation was set to zeros. Missing values could create problem in convergence of the model, so, zeros were inserted instead of the missing values.
Following methods were investigated in PARAFAC modeling of the data.
1. Hard Weighting
2. MILES Weighting
3. Soft Weighting
4. Constrained Hard Weights
5. Constrained MILES
The results showed MILES was the best technique for the data sets used to study PARAFAC modeling. So, MILES can be considered best among the techniques quoted above.
Contribution to ESEL
This paper discusses five techniques to handle Rayleigh scatter and the technique MILES gave best results. Inspite of the fact, we must check all the possibilities. It is because of the fact that PARAFAC modeling results vary with samples.
By: Aamir Alaud-din