Abstract in English: Deoxynivalenol (DON) is the most common Fusarium-mycotoxin among the type B group of
trichothecenes found at high concentration in cereals mainly produced by F. graminearum and F.
culmorum. Analytical methods for the rapid evaluation of mycotoxins contamination of grains are
highly needed in order to prevent the food chain contamination. The synthesis of mycotoxins is
often associated with the production of volatile organic compounds (VOCs). These, when
detectable through the use of suitable methods of analysis, represent a “fingerprint” of the
contaminated samples. PTR-TOFMS technology was used for a rapid and efficient detection of
DON in naturally contaminated durum wheat grain samples, by evaluating the correlations between
selected patterns of volatile organic compounds with DON concentration. A partial least square
(PLS) regression was used to efficiently quantify the DON value (r= 0.94 in the test set). A partial
least squares discriminant analysis (PLSDA) was used to classify the samples above or below the
fixed DON value of 1,750 ?g/kg (91.9% of correct classification in the test set). Eight VOCs have
been proven as good indicators in discriminating wheat samples, with DON concentration values
above/below the legal limit 1,750 ?g/kg, using both class-modeling approaches, partial least square
(PLS) analysis and partial least squares discriminant analysis (PLSDA). The obtained results show
a promising use of the PTR-TOFMS for the application of rapid, non-destructive, and massive
screening of DON contaminated durum wheat samples. |