67C-20 |
|
L. L. VINES1, S. E. Kays1, and P. E. Koehler2. (1) Quality Assessment Research Unit, USDA-ARS-Richard B. Russell Research Center, 950 College Station Rd., PO Box 5677, Athens, GA 30604-5677, (2) Dept. of Food Science & Technology, Univ. of Georgia, 317 Food Science Bldg., Athens, GA 30602-7610 AOAC method 996.01 is utilized to analyze total fat content in cereal foods, as defined by the Nutrition Labeling and Education Act (NLEA). Although accurate, the method is arduous and requires disposal of hazardous chemicals. In contrast, near-infrared (NIR) spectroscopy is a fast, accurate, and environmentally benign analytical technique. It was hypothesized that NIR reflectance spectroscopy can be used to predict total fat content in cereal products rapidly and accurately when AOAC Method 996.01 is the reference. The objective was to construct a NIR reflectance model for the prediction of total fat in cereal products using AOAC Method 996.01 as the reference method and analyze the model’s applicability for nutrition labeling. Cereal product samples with a range of fat contents, grains, and processing methods were purchased from retailers and ground. Reflectance spectra were obtained with a NIR dispersive grating spectrometer and total fat reference values determined by AOAC Method 996.01. Using a commercial multivariate analysis program, a modified partial least squares model (n=72) was developed to predict total fat content by relating reference values to NIR spectra (wavelength range 1104-2494 nm). Model accuracy was tested using independent validation samples of cereal products (n=35). A rapid and accurate NIR model for the prediction of total fat was obtained. The model had a standard error of cross validation of 1.12% (range 0.50-43.20%) and R2 of 0.99. When the model was used to predict independent validation samples, the resultant standard error of prediction was 0.97% total fat (range 2.10-35.70%) and r2 0.99 with a bias of 0.07%. Ninety-seven percent of the validation samples were predicted within NLEA accuracy guidelines. A NIR reflectance model for the rapid and accurate prediction of total fat in cereal foods has utility for the food industry, commercial analysis laboratories, and regulatory agencies that monitor labeling accuracy.
Session 67C, Food Chemistry: Food analysis, irradiation and toxicology
|