Mid-infrared diffuse reflectance spectroscopy for discriminant analysis of food ingredients

The objective was to determine if mid-infrared diffuse reflectance spectroscopy could be used to discriminate various food ingredients. Samples (n = 106) consisting of buttermilk, cheese, dehydrated onion and milk-egg powders, wheat flours, and two powdered seasonings (cheese and ranch) were scanned as neat powders using diffuse reflectance on Digilab FTS-60 and Perkin-Elmer model 2000 Fourier transform spectrometers. Discriminant analysis, using mean centering and multiplicative scatter-corrected spectra, was performed using Mahalanobis distance by principal component analysis using the averaged predicted distance or F-test indicator to select factors obtained from a one-out cross-validation analysis. Results from the two spectrometers demonstrated that discriminant analysis of food ingredient powders based on spectra of neat powders can be successfully carried out. In general, it was found that results using 4-cm(-1) resolution spectra were somewhat superior to those based on lower 16-cm(-1) resolution spectra.