Rapid detection of fumonisin B1 and B2 in ground corn samples using smartphone-controlled portable near-infrared spectrometry and chemometrics.

• Smartphone-controlled NIR spectrometer was successfully used to detect FBs contamination in corn. • The spectral absorbance intensity of corn samples varied with the different levels of FBs contamination. • LPLS-S showed higher prediction accuracy of FBs content in corn than PLS and SVM models. • PLS-DA and SVM-DA showed similar results for classifying corn samples according to the MRL of FBs. A portable near-infrared (NIR) spectrometer coupled with chemometrics for the detection of fumonisin B 1 and B 2 (FBs) in ground corn samples was proposed in the present work. A total of 173 corn samples were collected, and their FB contents were determined by HPLC–MS/MS. Partial least squares (PLS), support vector machine (SVM) and local PLS based on global PLS score (LPLS-S) algorithms were employed to construct quantitative models. The performance of the SVM and LPLS-S was better than that of PLS, and the LPLS-S presented the lowest RMSEP (12.08 mg/kg) and the highest RPD (3.44). Partial least squares-discriminant analysis (PLS-DA) and support vector machine-discriminant analysis (SVM-DA) were used to classify corn samples according to the maximum residue limit (MRL) of FBs, and the discriminant accuracy of both the PLS-DA and SVM-DA algorithms was above 86.0%. Thus, the present study provided a rapid method for monitoring FB contamination in corn samples. [ABSTRACT FROM AUTHOR]