Skip to main content
ARS Home » Research » Publications at this Location » Publication #181663


item Kim, Yookyung
item Kays, Sandra
item Singh, Mukti

Submitted to: Eastern Analytical Symposium
Publication Type: Abstract Only
Publication Acceptance Date: 5/17/2005
Publication Date: 11/15/2005
Citation: Kim, Y., Kays, S.E., Singh, M. 2005. Nir and ft-nir spectroscopy for the measurement of total dietary fiber in mixed foods. Eastern Analytical Symposium. Paper number 554.

Interpretive Summary:

Technical Abstract: Traditional methods of analysis of dietary fiber in mixed foods are very time consuming (taking 4-5 days), labor intensive, expensive, generate chemical waste and require extensive sample preparation. NIR spectroscopy has been used to accurately predict total dietary fiber (TDF) in cereal products. The aim of this study was to evaluate the feasibility of NIR spectroscopy for prediction of TDF in mixed foods using three levels of sample pretreatment: asis, dried, and dry/defatted. Samples were selected from retailers covering a wide range of TDF. Frozen, packaged, or canned meals were homogenized and NIR spectra obtained with dispersive and Fourier transform NIR spectrometers. TDF was measured by AOAC method 991.43 in dry/defatted samples. Based on fat and moisture loss, reference values were calculated for asis and dried samples. Using NIR spectra (700 to 2496 nm) of asis, dried, and dry/defatted samples and corresponding reference values for TDF, partial least squares (PLS) regression models were developed for prediction of TDF in mixed foods at the three levels of sample pretreatment. The PLS models were evaluated by cross validation and by prediction of an independent validation set. Using the dispersive NIR spectrometer, the SECV and R2 of the asis, dried, and dry/defatted PLS models developed (n=115) were 0.99 and 0.84, 1.83 and 0.93, and 1.54 and 0.96 respectively. The asis, dried, and dry/defatted models predicted TDF in independent validation samples (n=39) with SEP of 0.93, 1.90, and 1.45, respectively; r2 of 0.89, 0.92 and 0.97, respectively; and RPD of 2.04, 2.53 and 4.05, respectively. In comparison, FT-NIR spectra gave higher values of SECV and lower R2 using optimum pre-processing of spectra. NIRS provides a rapid method for screening TDF in mixed foods when samples are scanned after homogenizing only. Accuracy of TDF prediction increases with drying or drying and defatting samples before obtaining NIR spectra.