Abstract

Fourier-transform-near infrared (FT-NIR) spectroscopy has been used to develop quantitative and classification models for the prediction of deoxynivalenol (DON) levels in durum wheat samples. Partial least-squares (PLS) regression analysis was used to determine DON in wheat samples in the range of <50-16,000 g/kg DON. The model displayed a large root mean square error of prediction value (1,977 g/kg) as compared to the EU maximum limit for DON in unprocessed durum wheat (i.e., 1,750 g/kg), thus making the PLS approach unsuitable for quantitative prediction of DON in durum wheat. Linear discriminant analysis (LDA) was successfully used to differentiate wheat samples based on their DON content. A first approach used LDA to group wheat samples into three classes: A (DON <= 1,000 g/kg), B (1,000 < DON <= 2,500 g/kg), and C (DON > 2,500 g/kg) (LDA I). A second approach was used to discriminate highly contaminated wheat samples based on three different cut-off limits, namely 1,000 (LDA II), 1,200 (LDA III) and 1,400 g/kg DON (LDA IV). The overall classification and false compliant rates for the three models were 75%-90% and 3%-7%, respectively, with model LDA IV using a cut-off of 1,400 g/kg fulfilling the requirement of the European official guidelines for screening methods. These findings confirmed the suitability of FT-NIR to screen a large number of wheat samples for DON contamination and to verify the compliance with EU regulation.

Autore Pugliese

• De Girolamo Annalisa , Pascale Michelangelo , Cervellieri Salvatore

Tutti gli autori

• De Girolamo A.; Cervellieri S.; Visconti A.; Pascale M.

Titolo volume/Rivista

Toxins

Anno di pubblicazione

2014

ISSN

2072-6651

ISBN

Non Disponibile

Numero di citazioni Wos

Nessuna citazione

Ultimo Aggiornamento Citazioni

Non Disponibile

Numero di citazioni Scopus

Non Disponibile

Ultimo Aggiornamento Citazioni

Non Disponibile

Settori ERC

Non Disponibile

Codici ASJC

Non Disponibile