Near-infrared hyperspectral imaging as a novel approach for T-2 and HT-2 toxins estimation in oat samples

Oat and oat-based products are expected to present the highest levels of some Fusarium mycotoxins, such as T-2 and HT-2 toxins. The content of these trichothecenes has been increasing in oats in the last few years. Moreover, oat and oat-based products consumption is expanding, so it is even more relevant to control their mycotoxin content at the reception of the food and feed companies. Near-infrared hyperspectral imaging (NIR-HSI) is considered a promising technique for determining mycotoxins in cereals because it is quicker, cheaper and more sustainable than conventional techniques. However, it has never been used to assess T-2 and HT-2 toxins content. In the present study, NIR-HSI has been applied in 119 oat naturally contaminated samples to quantify T-2+HT-2 toxins and classify samples according to the EU indicative level for unprocessed oats (1000 μg/kg). Hyperspectral images were taken for unground and ground samples and the sum of T-2 and HT-2 toxins content was analysed by ELISA as the reference method. The data were pre-treated and analysed by multivariate tools, applying PLS regression for quantification and four classification models. The best-fitted T-2+HT-2 toxins prediction model was for unground samples and SNV plus 1st derivative pre-treatment (R2 = 0.64 and RMSECV = 157.80 μg/kg). The correlation between the predicted and observed T-2+HT-2 toxins values was 0.80, and the most characteristic wavelengths were 1038, 1110 and 1393 nm. The maximum classification accuracy was 94.1% for whole and milled samples according to the EU indicative level. These findings suggest that T-2 and HT-2 toxins detection by NIR-HSI is possible in oat grains, even if further investigations are needed.

This work was supported by the Spanish Ministry of Science and Innovation (predoctoral grant FPU21/00073 and Project PID 2020-114836RB-I00 funded by MCIN/AEI/10.13039/501100011033 ).