Classification of cereal flour species using Raman spectroscopy in combination with spectra quality control and multivariate statistical analysis

• Published in: Journal of cereal science Vol. 101; p. 103299
• Main Authors: Kniese, Jasmin, Race, Alan M., Schmidt, Heinar
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.09.2021
• Subjects: adulterated products; arabinoxylan; Authenticity; barley; discriminant analysis; flour; Harvest year; least squares; Mahalanobis distance; PCA; PLS-DA; Prediction; principal component analysis; quality control; Raman spectroscopy; rye; SIMCA; staple foods; starch; wheat; Authenticity; SIMCA; Prediction; Harvest year; PLS-DA; Mahalanobis distance; PCA
• ISSN: 0733-5210; 1095-9963
• DOI: 10.1016/j.jcs.2021.103299

As an important staple food, grain is subject to substitution and adulteration with cheaper cereals. Therefore, verification of authenticity is an issue for flour. In this work, Raman spectroscopy was used as non-invasive technique to discriminate the species barley, rye, spelt and wheat. In total, 129 samples with different origins and varieties from two harvest years were used as training data. Principal component analysis and partial least-squares discriminant analysis showed a clear differentiation of the species (accuracies >98 %). Quality control of anomalous spectra improved the classification accuracy with Mahalanobis distance and Hotelling's T2 and Q residuals statistics proving superior to manual selection. Validation with 86 independent samples including an additional harvest year per species corroborated the feasibility of discrimination (accuracy: 88 %) and shows the impact of the variety and harvest year. The discrimination of the spectra was primarily based on starch, protein and arabinoxylan signals. [Display omitted] •Classification of four grain flour species based on non-invasive Raman spectra.•Sampling over five harvest years showed impact of species, variety and year.•Discriminant model calibrated 98 % accuracy; validated 88 % accuracy.•Extreme weather conditions such as drought had an effect.•Discrimination was based on differences in starch, protein and arabinoxylan signals.