Skip to main content
ARS Home » Research » Publications at this Location » Publication #159611


item Kays, Sandra
item Archibald, Douglas
item Sohn, Mi Ryeong

Submitted to: Journal of the Science of Food and Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/10/2004
Publication Date: 3/31/2005
Citation: Kays, S.E., Archibald, D., Sohn, M. 2005. Prediction of fat content in intact cereal food products using NIR reflectance spectroscopy. Journal of the Science of Food and Agriculture. 85(9):1596-1602.

Interpretive Summary: Near infrared (NIR) reflectance spectroscopy is a technique that measures the amount of light energy reflected by a substance and relates that light energy to a measured component of the substance by mathematical modeling. The technique is rapid and accurate and does not require the use of chemicals or generate chemical waste. Fat content in foods is measured by standard techniques that use organic solvents requiring specific disposal. The techniques are often time consuming. All work on NIR determination of cereal product fat composition, so far, has been conducted on ground or milled products. It would be advantageous to the manufacturer and food monitoring agencies to be able to determine fat content directly without the need for grinding and the extra time involved. The objective of this study was to determine the potential of NIR spectroscopy for the determination of fat content in a diverse range of intact cereal products (including breakfast cereals, crackers and cereal based snacks). An NIR model was developed for the prediction of crude fat content in diverse cereals. The model is accurate enough for rough screening in an on-line situation and could be used by manufacturers and monitoring agencies to screen for products that are not in line with nutrition label values.

Technical Abstract: The potential of NIR reflectance spectroscopy for the quantitative analysis of crude fat in intact cereal products was determined. Reflectance spectra (400-1700 nm) of intact cereal foods were obtained using a diode array spectrometer. Fat content was analyzed gravimetrically, following extraction with petroleum ether (AOAC Method 945.16). Using multivariate analysis, partial least squares models were built to predict the fat content of independent validation samples (n=52) with sufficient accuracy for screening samples. The standard error of performance (SEP) and multiple coefficient of determination were 1.61-1.65% fat and 0.95-0.96, respectively. The model was expanded to include samples with a broad range of particle sizes and moisture contents without reduction in accuracy of prediction (SEP of 1.13% and multiple coefficient of determination of 0.95). Model development using selection of the best wavelength regions by Marten's Uncertainty Test, prior to PLS regression, resulted in models with fewer factors. Wavelength regions at 1215 and 1390 nm appeared to be the most valuable for model development. Keywords: near infrared, NIR, cereal, fat.