Utilisation of visible/near-infrared hyperspectral images to classify aflatoxin B1 contaminated maize kernels

Authors: KIMULI, D - China Agricultural University; WANG, W - China Agricultural University; Lawrence, Kurt; Yoon, Seung-Chul; Ni, Xinzhi; Heitschmidt, Gerald - Jerry

Submitted to: Biosystems Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/27/2017
Publication Date: 1/15/2018
Citation: Kimuli, D., Wang, W., Lawrence, K.C., Yoon, S.C., Ni, X., Heitschmidt, G.W. 2018. Utilisation of visible/near-infrared hyperspectral images to classify aflatoxin B1 contaminated maize kernels. Biosystems Engineering. 166:150-160.

Interpretive Summary: A visible/near-infrared hyperspectral imaging system with a wavelength range between 400 and 1000 nm was used to determine the feasibility of detecting aflatoxin on surfaces of maize kernels collected from four maize varieties from different regions of the United States of America, i.e. Georgia, Illinois, Indiana and Nebraska. For each variety, four aflatoxin concentrations (10, 20, 100 and 500 ppb) were artificially applied on maize kernel surfaces. Similarly, a control group was generated from 30 kernels of each variety treated with a solution of methanol. Principal component analysis and factorial discriminant analysis were utilized to determine the pattern of separation between uncontaminated and contaminated kernels. Factorial discriminant analysis showed the ability to predict aflatoxin contamination of each variety with 96% validation accuracy while prediction for aflatoxin contamination group membership of pooled samples reached 98% accuracy in validation. Variation in the spectra of aflatoxin contaminated kernels might have caused the variation in the predicted aflatoxin contamination group membership. This study presents the potential of using visible/near-infrared hyperspectral imaging system in classifying aflatoxin contamination of maize kernels in different varieties. The current study further suggests that varietal differences of maize kernels may have no influence on aflatoxin contamination classification.

Technical Abstract: A visible/near-infrared (VNIR) hyperspectral imaging (HIS) system (400 and 1000 nm) was used to assess the feasibility of detecting aflatoxin B1 (AFB1) on surfaces of four maize varieties from different regions of the United States of America, i.e. Georgia, Illinois, Indiana and Nebraska. For each variety, four AFB1 solutions (10, 20, 100 and 500 ppb) were artificially applied on kernel surfaces. Similarly, a control group was generated from 30 kernels of each variety treated with a solution of methanol. Principal component analysis (PCA) was used to reduce dimensionality of the hyperspectral image data followed by the application of factorial discriminant analysis (FDA) on the principal component variables. PCA results showed a pattern of separation between uncontaminated and contaminated kernels for all varieties except for Indiana samples and pooled samples. FDA showed the ability to predict AFB1 contamination of each variety with 96% validation accuracy while prediction for AFB1 contamination group membership of pooled samples reached 98% accuracy in validation. Variation in the spectra of AFB1 contaminated kernels could have caused the variation in the predicted AFB1 contamination group membership. The PCA and FDA models where influenced by the chemical information from C-H, N-H and O-H bonds of Vis/NIR spectral regions. This study presents the potential of using VNIR hyperspectral imaging in direct AFB1 contamination classification studies of maize kernels of different varieties. The current study further suggests that varietal differences of maize kernels may have no influence on AFB1 contamination classification.