A novel chemometric classification for FTIR spectra of mycotoxin-contaminated maize and peanuts at regulatory limits

Abstract

The rapid identification of mycotoxins such as deoxynivalenol and aflatoxin B in agricultural commodities is an ongoing concern for food importers and processors. While sophisticated chromatography-based methods are well established for regulatory testing by food safety authorities, few techniques exist to provide a rapid assessment for traders. This study advances the development of a mid-infrared spectroscopic method, recording spectra with little sample preparation. Spectral data were classified using a bootstrap-aggregated (bagged) decision tree method, evaluating the protein and carbohydrate absorption regions of the spectrum. The method was able to classify 79% of 110 maize samples at the European Union regulatory limit for deoxynivalenol of 1750 µg kg and, for the first time, 77% of 92 peanut samples at 8 µg kg of aflatoxin B. A subset model revealed a dependency on variety and type of fungal infection. The employed CRC and SBL maize varieties could be pooled in the model with a reduction of classification accuracy from 90% to 79%. Samples infected with Fusarium verticillioides were removed, leaving samples infected with F. graminearum and F. culmorum in the dataset improving classification accuracy from 73% to 79%. A 500 µg kg classification threshold for deoxynivalenol in maize performed even better with 85% accuracy. This is assumed to be due to a larger number of samples around the threshold increasing representativity. Comparison with established principal component analysis classification, which consistently showed overlapping clusters, confirmed the superior performance of bagged decision tree classification.

Fourier transform near‐infrared and mid‐infrared spectroscopy as efficient tools for rapid screening of deoxynivalenol contamination in wheat bran

“…). These results were in agreement with those recently reported by Kos et al . who proposed a bootstrap‐aggregated (bagged) decision‐tree classification approach for the screening of maize samples contaminated with DON at levels up to 43 500 µg kg −1 .…”

“…3). These results were in agreement with those recently reported by Kos et al 26 who proposed a bootstrap-aggregated (bagged) decision-tree classification approach for the screening of maize samples contaminated with DON at levels up to 43 500 μg kg −1 . The FTMIR model was able to classify 85% of samples correctly with 1% of FC samples when a 500 μg kg −1 cutoff limit was used to discriminate the two classes of maize samples.…”

“…The FTMIR model was able to classify 85% of samples correctly with 1% of FC samples when a 500 µg kg −1 cutoff limit was used to discriminate the two classes of maize samples. However, a sieving step before ATR measurements and the use of the fraction with a particle diameter between 250 and 100 µm was needed to improve the repeatability of the spectral measurements significantly . On the other hand, the milling procedure used in the present paper guaranteed the use of the entire sample (by eliminating the sieving step) and gave an acceptable homogeneity of particle sizes required for the ATR measurements.…”

Gallium arsenide waveguides as a platform for direct mid-infrared vibrational spectroscopy

“…While the direct detection of mycotoxins may be limited with ATR-based IR sensors [30][31][32], the evaluation of matrix changes resulting from fungal activity after infection during growth, harvest, or storage scenarios may be promising. As previously shown, matrix changes are indeed correlated with the presence of mycotoxins, as shown via FT-IR-and QCL-based techniques with sensitivities at EU regulatory limits [33][34][35][36][37][38]. However, utilizing broadband spectra of selected mycotoxins along with smart extraction schemes may increase the utility of MIR monitoring systems, as shown for the example of AFB1.…”

Updated Overview of Infrared Spectroscopy Methods for Detecting Mycotoxins on Cereals (Corn, Wheat, and Barley)

“…DON assay quantification models are poor at predicting a specific grade, especially with respect to regulatory grades. However, these models can be used to classify grain lots according to a level of toxin contamination [ 40 , 41 , 49 , 64 , 66 ]. Peiris et al (2009, 2010) [ 39 , 40 ] used a grain-to-grain method to identify wheat grains infected with Fusarium and to predict DON levels.…”