Detection of Fusarium head blight contamination in wheat kernels by multivariate imaging

Abstract

The objective of this study was to evaluate the safety quality of wheat kernels and more specifically the infection of wheat by Fusarium culmorum. We have developed a new non-destructive methodology based on multispectral imaging and chemometrics to detect wheat contamination. This method has been applied to evaluate the tolerance to Fusarium Head Blight (FHB) on six accessions of durum wheat. Two of the accessions are parental lines, one susceptible and one resistant to Fusarium, and the other four are their offspring lines.Multispectral images of dorsal and ventral sides of all kernels were acquired with an in-house imaging system and processed with chemometrical tools, including principal component analysis (PCA) and multiple linear regression (MLR). According to the quantitative PCR (qPCR) results, PCA was able to detect infested kernels and to identify affected areas with kernels. Moreover, regression analysis of the images allowed the degree of contamination for each pixel to be estimated, and thus mapping of contamination for each kernel.