Author
 |
Archibald, Douglas |
|
Kays, Sandra |
|
Himmelsbach, David |
|
Barton Ii, Franklin |
|
Submitted to: Journal of Applied Spectroscopy
Publication Type: Peer Reviewed Journal Publication Acceptance Date: 7/5/1997 Publication Date: N/A Citation: N/A Interpretive Summary: This work studies the use of multiwavelength light reflection to determine the dietary fiber content of a wide range of cereal food products. The main goal is to replace the current wet chemical method which is laborious and wasteful. The technical approach involves measuring light reflection from a representative set of cereal food samples with known dietary fiber content and subsequently developing a mathematical calibration model to allow prediction of dietary fiber from the optical information. For these purposes, 3 types of mathematical methods and 2 types of light interaction were studied; the optical methods were near-infrared-reflectance and Raman scattering. This paper reports evaluation of the 6 resulting models using a new set of cereal samples with known dietary fiber values and a detailed list of additional attributes such as cereal grain type and fat or sugar content. The findings suggest that improved determination methods would be eobtained if certain sample subgroups are better represented in the data se and if two or more different mathematical and optical methods are combined. The Raman method appears slightly less precise than the near-infrared method, but tolerates a wider range of cereal food types. The study also identified several potential ways to improve the Raman experimental technique for agricultural and food applications. Technical Abstract: This work evaluates the complementarity in the predictive ability of three Raman and three near-infrared reflectance (NIRR) partial least squares regression (PLSR) models for total dietary fiber (TDF) determinations of a diverse set of ground cereal food products. For each spectral type (R or N), models had previously been developed from smoothed (DO), first derivative (D1), or second derivative (D2) spectral data. The NIRR and Raman models tend to have very different sets of outliers and uncorrelated errors in TDF determination. For a single spectral type, the prediction errors of various preprocessing methods are partially complementary. The samples are very diverse in terms of composition, but main problem groups were narrowed to: high-fat, high-bran, high-germ, and those containing synthetic fiber additives. Raman models perform better on the high-fat samples, while NIRR models perform better with high-bran and high-synthetic csamples. Raman models were better able to accommodate a wheat germ sample even though this was poorly represented by the calibration set. Two methods are presented for utilizing the complementarity of the spectral and processing techniques: one involves simple averaging of predictions; and the other involves avoidance of outliers by using statistics generated from the sample spectrum to choose the best model(s) for determination of the TDF value. The single best model (ND1) has a root-mean-squared prediction error of 2.4% TDF. The best model of prediction averages yields an error of 1.9% (combining ND0, ND1, ND2, RD0 and RD1). An error of 1.9% was also obtained by choosing a single prediction from the six models by using statistics to avoid outliers. By choosing the best three models and averaging their predictions, they achieved an error of 1.5%. |
