Skip to main content
ARS Home » Research » Publications at this Location » Publication #84139

Title: DISCRIMINATE ANALYSIS OF FOOD INGREDIENTS BY NEAR-INFRARED DIFFUSE REFLECTANCE SPECTROSCOPY.

Author
item Reeves Iii, James
item ZAPF, CHARLES - MCCORMICK

Submitted to: Near Infrared Spectroscopy Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/30/1997
Publication Date: N/A
Citation: N/A

Interpretive Summary: Spectroscopy is the study of how light interacts with materials. By using light of specific types (i.e., blue versus red) and examining the interaction with different materials it is often possible to determine the composition of the material in question, or to classify the material (discriminate analysis). Near-infrared which is light just beyond the ability of humans to see (some of what we feel as heat) has been extensively used over the last twenty years to determine the composition of materials such as animal feeds. Discriminate analysis, on the other hand, is an effort to determine what something is or is not. For proper discrimination a product, such as wheat flour, needs to be classified as being a wheat flour and not something else. The objective of this work was to determine the potential of using near-infrared spectra (graphical representation of interaction between the specific light and a given material) to discriminate among food ingredients. For this work, 106 samples consisting of buttermilk, dehydrated onion, cheese and milk-egg powders, wheat flours, and two powdered seasonings were examined in the near-infrared. In general, it was found that correct classifications could be achieved. However, trial and error adjustments were often required for the best results.

Technical Abstract: The objective of this research was to determine if near-infrared (NIR) diffuse reflectance spectroscopy could be used to discriminate among various food ingredients, and the influence of the type of instrumentation used. Samples (106) consisting of milk-egg, buttermilk, cheese and dehydrated onion powders, wheat flours, cheese and ranch seasonings were scanned using diffuse reflectance on a NIRSystems model 6500 monochromator and a Digi-Lab FTS-60 Fourier transform. With the monochromator, samples were scanned (32 co-added scans) using a rotating sample cup from 1100 to 2498 nm (9091 to 4003 cm-1) with data collected every 2 nm at a resolution of 10 nm. For the FTS-60, a linear motion sample transport was used with 64 co-added scans at 4 and 16 cm-1 resolutions. Analysis was performed using "Mahalanobis Distance by PCA with Residuals" using the "Averaged Predicted Distance" indicator to determine the number of factors, obtained from a one-out cross validation analysis, to use for each grouping, with either all available samples or with one third of the samples reserved as an independent test set. All spectra were pretreated using mean centering and multiplicative scatter correction, with and without derivatization. Using all available samples, calibrations could be developed using either spectrometer which were capable of discriminating among the various groups of food ingredients. For the FTS-60, results were considerably better for the 4 cm-1 resolution spectra (compared to the 16 cm-1) when samples were divided into calibration and test sets. In conclusion, these results demonstrate that NIR spectra of food ingredients obtained from either a scanning monochromator or a Fourier transform spectrometer can be used for product discrimination or identification.