Near infrared spectroscopy for determination of mycotoxins in cereals

The use of near infrared transmittance instrumentation for determination of the mycotoxin deoxynivalenol (DON) in wheat kernel samples was investigated in a Nordic pilot project. Dilution series of wheat field infected with Fusarium culmorum was used in the calibration trials. Principal component analysis and partial least square regression calculations showed that the number of factors required to predict DON was 11-13 depending on wavelength range used. The best regression model was obtained for the wavelength range 670-1100 nm with a slope of 0.949, a correlation coefficient of 0.984 and a standard error of 381 mug DON per kg. The results indicate that it may be possible in the future to develop a calibration model, which can be used to screen for DON in wheat at concentrations just above the proposed European limits for wheat flour. (C) 2003 Elsevier Science Ltd. All rights reserved.