Non-destructive measurement of deoxynivalenol in wheat flour using fluorescence fingerprinting

Abstract

A method of predicting deoxynivalenol (DON) concentrations in wheat flour was developed using fluorescence fingerprinting (FF), which is also known as excitation-emission matrix. A partial least square (PLS) regression model was built based on FF values at various wavelength conditions and reference values for DON concentrations were measured by chemical analysis. The model showed good fitting to the validation dataset with R2=0.983 and SEP = 1.4 ppm. The absolute values of the regression coefficient for the calibration model were significantly high in the wavelength range of Ex 240-400 nm and Em 500-600 nm. To reduce the number of wavelength conditions for shorter measurement time, PLS regression was again applied to the FF values at the above limited wavelength conditions. The fitting values of the acquired model to the validation dataset were R2 = 0.977, SEP = 1.6 ppm, and were not significantly different from those of the first model. The results indicate that rapid and non-destructive prediction of DON concentration in wheat flour is possible using FF.