Skip to main content
ARS Home » Southeast Area » New Orleans, Louisiana » Southern Regional Research Center » Food and Feed Safety Research » Research » Publications at this Location » Publication #316992

Title: Integration of fluorescence and reflectance visible near-infrared (VNIR) hyperspectral images for detection of aflatoxins in corn kernels

Author
item ZHU, FENGLE - MISSISSIPPI STATE UNIVERSITY
item YAO, HAIBO - MISSISSIPPI STATE UNIVERSITY
item HRUSKA, ZUZANA - MISSISSIPPI STATE UNIVERSITY
item KINCAID, RUSSELL - MISSISSIPPI STATE UNIVERSITY
item BROWN, ROBERT
item Bhatnagar, Deepak
item CLEVELAND, THOMAS

Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/23/2016
Publication Date: 3/25/2016
Citation: Zhu, F., Yao, H., Hruska, Z., Kincaid, R., Brown, R.L., Bhatnagar, D., Cleveland, T.E. 2016. Integration of fluorescence and reflectance visible near-infrared (VNIR) hyperspectral images for detection of aflatoxins in corn kernels. Transactions of the ASABE. 59(3):785-794. https://doi.org/10.13031/trans.59.11365.
DOI: https://doi.org/10.13031/trans.59.11365

Interpretive Summary: Aflatoxins are poisons produced by the fungus Aspergillus flavus after it infects agricultural commodities such as corn. Since aflatoxins in food and feed are regulated, enhanced ability to detect and measure fungal growth and aflatoxin contamination of corn could contribute significantly towards the separation of contaminated from healthy grain. A collaboration between ARS-SRRC, Food and Feed Safety Research Unit and Mississippi State University, Stennis Space Center, MS is exploring the use of hyperspectral imaging non-destructive technology (developed by ITD) to detect mycotoxin-producing fungi and mycotoxins in grain products. The present experiment compared spectral data of corn infected in the field with a toxigenic Aspergillus flavus fungal strain obtained by one of three imaging methods: fluorescence, reflectance, or integration of the two. Results showed that the most accurate hyperspectral image data came from the integrated analysis. Further experiments may lead to this technology and approach being used to rapidly and accurately detect/measure Aspergillus flavus infection/aflatoxin contamination of corn without destruction of healthy grain. This could provide a useful tool to both growers and buyers in the corn industry that could enhance protection of food and feed as well as increase profits.

Technical Abstract: Aflatoxin contamination in agricultural products has been an important and long-standing problem around the world. Produced by certain fungal species of the Aspergillus genus, aflatoxins are highly toxic and carcinogenic. This study investigated the integration of fluorescence and reflectance visible near-infrared (VNIR) hyperspectral images to detect aflatoxins in whole corn kernels. Artificial inoculation on corn ears was implemented with toxigenic A. flavus spores at the early dough stage in the field. After harvest, three groups of kernels were collected from the inoculated ears visually inspected under ultraviolet (UV) light: fluorescent glowing, adjacent to glowing, and controls. Both fluorescence hyperspectral images under UV excitation and the reflectance hyperspectral images under halogen illumination were recorded on the two sides of kernels (endosperm and germ). Subsequent chemical analysis was performed on each kernel sample to provide reference aflatoxin concentration. Threshold values of 20 ppb and 100 ppb were adopted separately to group kernels into contaminated and healthy. In comparison with healthy kernels, fluorescence spectral curves of contaminated kernels had peaks shifted to longer wavelengths with lower intensity, and reflectance values of contaminated kernels were generally lower with a different spectra-changing trend in the 700-800 nm range. Spectral datasets were compressed and interpreted using principal component analysis (PCA). Least squares support vector machines (LS-SVM) and k-nearest neighbor (KNN) classifiers were performed on the fluorescence PC variables, reflectance PC variables, integrated fluorescence and reflectance PC variables, respectively for classifying contaminated and healthy kernels on both sides of kernels. The best overall prediction accuracy was 95.33% from the LS-SVM model for the 100 ppb threshold on germ side in the integrated data analysis. Overall, the germ side performed better than the endosperm side, especially for the true positive rate (TPR). Fluorescence and reflectance image data generally achieved similar classification accuracy. Their integrated analysis achieved better results than single fluorescence or reflectance analysis on the germ side. The TPR of germ side increased from an average of 84.02% in fluorescence analysis and 83.34% in reflectance analysis to an average of 90.01% in the integrated data analysis. The mean aflatoxin concentration in the prediction samples reduced from 2662.01 ppb to 64.04 ppb, 87.33 ppb, and 7.59 ppb, respectively, after removing samples that were classified as contaminated by fluorescence, reflectance, and integrated data analysis, respectively, on the germ side. Since fluorescence and reflectance measurements can be made with the same imaging system, the integrated technique applied to the germ side could provide better elimination of heavily contaminated kernels.