Hyperspectral imaging techniques for rapid detection of single- and co-contaminant aflatoxins and fumonisins in ground maize

Authors: KIM, YONG-KYUNG - National Agricultural Products Quality Management Service; Baek, Insuck; LEE, KYUNG-MIN - Texas A&M Agrilife; KIM, GEONWOO - Gyeongsang National University; KIM, SEYEON - National Agricultural Products Quality Management Service; KIM, SUNG-YOUN - National Agricultural Products Quality Management Service; Chan, Diane; HERRMAN, TIMOTHY - Texas A&M Agrilife; KIM, NAMSUK - National Agricultural Products Quality Management Service; Kim, Moon

Submitted to: Toxins
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/20/2023
Publication Date: 7/22/2023
Citation: Kim, Y., Baek, I., Lee, K., Kim, G., Kim, S., Kim, S., Chan, D.E., Herrman, T., Kim, N., Kim, M.S. 2023. Hyperspectral imaging techniques for rapid detection of single- and co-contaminant aflatoxins and fumonisins in ground maize. Toxins. 15(7):472. https://doi.org/10.3390/toxins15070472.
DOI: https://doi.org/10.3390/toxins15070472

Interpretive Summary: Aflatoxins and fumonisins are substances produced by some species of fungi and can contaminate corn (maize) and corn-derived products and subsequently cause severe illness in animals and humans if consumed in significant amounts. Methods to rapidly detect these fungal mycotoxins are needed for food safety screening of corn-based food ingredients. Established wet-chemistry methods work well but are time- and resource-consuming. Previous research to develop detection methods have focused primarily on detection of single contaminants, but these mycotoxins commonly occur as co-contaminants, present in the same product at the same time, and thus methods capable of simultaneously detecting multiple contaminants and doing so rapidly are needed. This study investigated hyperspectral imaging techniques using fluorescence, visible and near-infrared (VNIR) reflectance, and short-wave infrared (SWIR) reflectance to detect naturally occurring aflatoxin (AF) and fumosin (FM), both individually and together as co-contaminants, in samples of ground corn. Two types of classification models—Partial Least Squares-Discriminant Analysis (PLSDA), and Support Vector Machine (SVM) with radial basis function—were developed in combination with the use of different spectral preprocessing methods and tested for contaminant detection based on fluorescence, VNIR, and SWIR hyperspectral imaging. For samples of ground corn contaminated with both AF and FM, only AF, only FM, or no detectable levels of contaminants, the classification models demonstrated detection accuracies ranging between 70% and 96%, with the SVM model using SWIR imaging outperforming the other models. The results suggest that hyperspectral SWIR imaging may be more effective than VNIR or fluorescence in detecting AF and FM in corn, and could be a feasible method to implement as an inexpensive and user-friendly tool for food processors and distributors to ensure safe corn-based products for consumption by humans and animals.

Technical Abstract: Aflatoxins and fumonisins, commonly found in corn and corn-derived products, frequently co-occur and can cause severe illness in animals and humans if consumed in significant amounts. For early detection to prevent illness, efforts have been increased to develop analytical methods suitable for rapid mycotoxin screening. The applicability of hyperspectral imaging techniques, including reflectance in the visible and near-infrared (VNIR) and short-wave infrared (SWIR) regions, and fluorescence, was investigated for the classification of ground maize samples. The machine learning classification models in combination with different preprocessing methods were applied to screen ground maize samples for naturally occurring aflatoxin and fumonisin as single contaminants and as co-contaminants. The partial least squares-discriminant analysis (PLS-DA) and support vector machine (SVM) with radial basis function (RBF) kernel were employed as classification models using cut-off values of each mycotoxin. The classification performance of the SVM was better than that of PLS-DA, and the highest classification accuracies for fluorescence, VNIR, and SWIR were 89.1%, 71.7%, and 95.7% respectively. SWIR imaging with SVM model resulted in higher classification accuracies compared to the fluorescence and VNIR models, suggesting that as an alternative to conventional wet chemical methods, the hyperspectral SWIR imaging detection model may be the more effective and efficient analytical tool for mycotoxin analysis compared to fluorescence or VNIR imaging models. These methods show promise as inexpensive and easy-to-use food safety screening tools to rapidly detect mycotoxins in maize or other food ingredients intended for animal or human consumption.