Application of Raman spectroscopy for qualitative and quantitative analysis of aflatoxins in ground maize samples

• Raman spectroscopic technique was applied for aflatoxins detection in maize.
• A rapid and non-destructive spectroscopic method for aflatoxin was developed.
• Aflatoxin contaminated samples were not misclassified as negative by the model.
• The selected models for aflatoxin quantification showed high predictive accuracy.
• The method would help improve the safety of cereal-based food and feed products.

The applicability of Raman spectroscopy combined with chemometrics using different preprocessed spectra data was examined to develop fast, low-cost, and non-destructive spectroscopic methods for classification and quantification of aflatoxin-contaminated maize samples within the aflatoxin concentration range of 0–1206 μg/kg. This technique will find useful application in evaluating large numbers (e.g. >2000) of samples from maize hybrid performance trials and breeding programs. The best discriminant models were obtained from the linear discriminant analysis (LDA). The LDA models on validation samples showed correct classification rates in the range of 94–100% and did not misclassify any aflatoxin contaminated samples as aflatoxin negative. Of the models for predicting aflatoxin concentration, the partial least squares regression (PLSR) models showed the best quality of regression (slopes of 0.939–0.990) and highest coefficient of determination (r2 = 0.941–0.957). The models provide limited applicability to quantify aflatoxin concentration below 20 μg/kg. No significant difference was observed between predicted values using Raman spectroscopy and reference values using high-performance liquid chromatography (HPLC) (p > 0.05), indicating the suitability of Raman spectroscopy to rapidly screen large numbers of maize samples for aflatoxin contamination.