Identification of aflatoxin B1 on maize kernel surfaces using hyperspectral imaging

Abstract

A shortwave infrared (SWIR) hyperspectral imaging system with wavelength range between 1000 and 2500 nm was used to assess the potential to detect aflatoxin B1 (AFB1) contaminants on the surface of healthy maize kernels. Four different AFB1 solutions were prepared and deposited on kernels surface to achieve 10, 20, 100, and 500 ppb, respectively. A drop of 20% methanol was dipped on the surface of 30 healthy kernels in the same way to generate the control samples. Based on the standard normal variate (SNV) transformation spectra, principal components analysis (PCA) was used to reduce the dimensionality of the spectral data, and then stepwise factorial discriminant analysis (FDA) was performed on latent variables provided by the PCA's. Furthermore, beta coefficients of the first three of four discriminant factors were analyzed and key wavelengths, which can represent AFB1 and be used to differentiate different level of AFB1 were indentified. Furthermore, 150 independent samples were used as verification set to test the reproducibility of the proposed method. A minimum classification accuracy of 88% was achieved for the validation set and verification set. Results indicated that hyperspectral imaging technology, accompanied by the PCA-FDA method, can be used to detect AFB1 at concentrations as low as 10 ppb when applied directly on the maize surface.