Attenuated total reflection mid-infrared spectroscopy to screen Austrian and French wheat from multiple years for deoxynivalenol.

The high-throughput capability of attenuated total reflection mid-infrared spectroscopy (ATR-MIRS) makes it a promising analytical technique for rapid and green mycotoxin screening. In ATR-MIRS, changes in samples induced by mycotoxigenic fungi are correlated with mycotoxin concentrations obtained through reference analysis. Due to its indirect nature, limited research has explored the applicability of this technique for complex sample sets. We demonstrate that ATR-MIRS can effectively screen for deoxynivalenol (DON) contamination in wheat samples collected across two countries over multiple years. A total of 320 naturally contaminated samples from Austria and France were analyzed to develop screening models. Partial least squares discriminant analysis (PLS-DA) was used to classify samples as compliant or non-compliant with the European Commission (EC) limit of 1000 µg/kg DON in unprocessed wheat. Model performance during repeated nested cross-validation exhibited a true positive rate ranging from 0.32 to 1. This variability was primarily influenced by sample splitting, as well as by dataset balancing and spectral preprocessing approaches. These findings underscore the critical importance of sample selection when developing chemometric models for mycotoxin screening. Analysis of variable importance in projection (VIP) scores revealed that classification into compliant or non-compliant is based on wavenumbers associated with dissolved proteins and carbohydrates in the MIRS spectra. Overall, our results demonstrate the feasibility of using ATR-MIRS to assess DON contamination in complex, multiyear wheat sample sets while adhering to regulatory limits. Additionally, this study highlights the potential of MIRS for investigating the effects of mycotoxigenic fungi on wheat composition during the development of DON screening models.