Differentiating Rice Varieties by Inductively Coupled Plasma Mass Spectrometry Chemical Profiling with Singular Value Decomposition Background Correction

Abstract

As a global staple, rice has an increasing demand for differentiation and authentication, considering both fraud and health aspects for consumers. The complexity of rice urges a rapid, labor-saving, effective, and economical rice differentiation method. A semi-quantitative method using inductively coupled plasma mass spectrometry (ICP–MS) for fast and high-throughput analysis of rice entire elements with singular value decomposition (SVD) background correction (BC) was developed. The SVD BC method is the first time that it has been used with ICP–MS to minimize the effect of instrumental drift. Other data preprocessing methods, normalization (NM), error scaling (ES), and standard normal variate (SNV) were evaluated by principal component analysis (PCA) and projected difference resolution (PDR). A fuzzy rule-building expert system (FuRES), super partial least squares discriminant analysis (sPLS-DA), support vector machine (SVM), and SVM classification tree (SVMTreeG) classifiers gave 98.1 ± 0.0, 99.8 ± 0.4, 99.8 ± 0.4, and 100.0 ± 0.0% prediction rates with all of the data preprocessing methods (NM, SVD BC with one component to reconstruct the background, ES, and SNV) for the differentiation of the nine rice-digested samples, respectively.