Rapid and nondestructive quantification of deoxynivalenol in individual wheat kernels using near-infrared hyperspectral imaging and chemometrics

Abstract

The present study aimed to evaluate the feasibility of using near-infrared hyperspectral imaging (NIR-HSI) and chemometrics for quantifying deoxynivalenol (DON) in individual wheat kernels. In total, 120 wheat kernels of severely damaged kernels, moderately damaged kernels and asymptomatic kernels (SDKs, MDKs and AKs, respectively) were collected, and the DON content in the individual wheat kernels was analyzed by HPLC-MS/MS. Partial least squares (PLS), support vector machine (SVM) and local PLS based on global PLS scores (LPLS-S) algorithms were employed for building quantification models of DON. The results showed that SDKs and MDKs might contain low or no DON, while AKs could have a high DON content. Comparing the three modeling strategies, LPLS-S using mixed spectra achieved the best performance for kernels with RMSEP of 40.25 mg/kg and RPD of 2.24, which confirmed that NIR-HSI could be a feasible method for monitoring DON in individual kernels and removing highly contaminated kernels prior to food chain entry. • The impact of the kernel orientation (crease facing down or up) on the model results was investigated. • Wheat kernels were classified according to the degree of Fusarium infection by NIR-HSI and PLS-DA. • NIR-HSI combined with chemometrics was employed to quantify DON content in individual kernels.