Author
 |
Kays, Sandra |
|
Barton Ii, Franklin |
|
Submitted to: Dietary Fiber Conference Proceedings
Publication Type: Abstract Only Publication Acceptance Date: 2/18/2000 Publication Date: 5/13/2000 Citation: Kays, S.E., Barton II, F.E. 2000. Measurement of dietary fiber fractions by near infrared reflectance spectroscopy. Dietary Fiber Conference Proceedings. Abstracts p. 84. Interpretive Summary: Technical Abstract: The potential of using near infrared (NIR) reflectance spectroscopy for the rapid and accurate evaluation of soluble and insoluble dietary fiber was explored. Cereal products (breakfast cereals, crackers, pastas, brans, flours) with diverse grains and large variations in fat and sugar content were obtained from retailers, ground (<500 um) and analyzed for soluble and dinsoluble dietary fiber (AOAC Method 991.43). Ground samples were scanned with a scanning monochromator and NIR spectra collected and processed using a commercial analysis program. Using modified partial least squares regression, a NIR model was obtained (n=143) for the rapid and accurate determination of insoluble dietary fiber. The standard error of cross validation (SECV) was 1.5% and a multiple coefficient of determination 0.98. Prediction of an independent group of cereal samples (n=62) resulted in a standard error of performance and a coefficient of determination of 1.1% and 0.99, respectively. A model was also developed for the prediction of soluble dietary fiber in cereal products with low fat and sugar content. The model had a SECV of 1.1% and a multiple coefficient of determination 0.87. This model was expanded to include high fat and sugar and the resulting SECV and a coefficient of determination were 1.2% and 0.82, respectively. Prediction of validation samples by the soluble dietary fiber models resulted in a consistent bias, which may be related to differences in the way the references method treats samples with different soluble fiber constituents. |
