Detection of durum wheat pasta adulteration with common wheat by infrared spectroscopy and chemometrics: A case study

Abstract

Fourier transform (FT) infrared spectroscopy, in combination with Partial-Least Squares Discriminant Analysis (PLS-DA) and Linear Discriminant Analysis (LDA), was used to discriminate commercial durum wheat pasta from Italy and Argentina for common wheat adulteration. Samples were analyzed by both near- and mid-infrared spectroscopy (FT-NIR, FT-MIR) and the performance results were compared. Classification models were devel oped and validated using Argentinean and Italian durum wheat pasta samples containing common wheat at levels up to 28% and lower than 0.5%, respectively (as determined by ELISA method). The first LDA and PLS-DA models grouped samples into three-classes, i.e. common wheat ≤1%, from 1 to ≤5% and > 5%; while the second LDA and PLS-DA models grouped samples into two-classes using a cut-off of 2% common wheat. The accuracy of the validated models were between 80 and 95% for the three-classes approach and between 91 and 97% for the two-classes approach. In general, the three-classes approach provided better results in the FT-NIR range while the two-classes approach provided comparable results in both spectral ranges. Results indicate that FT-NIR and FT-MIR spectroscopy, in combination with chemometric models, represent a promising, inexpensive and easy-to-use screening tool to rapidly analyze durum wheat pasta samples for monitoring common wheat adulteration.