ABSTRACT
Soybean meal is a protein supplement for animals that may be treated with the adulterant melamine. This study reported the determination of melamine by near-infrared imaging. The near-infrared spectrum of pure soybean meal was used as the background to enhance the melamine spectrum. The melamine in adulterated samples was determined by partial least squares discriminant analysis for qualitative analysis. A linear relationship between melamine concentrations and the ratios of melamine pixels to all pixels was obtained. The results showed the presence of two linear correlations for lower and higher melamine concentrations. This method, designated as pixel number ratio-based concentration prediction, separately established low and high melamine concentration models for semiquantitative analysis. The method offers rapid analysis with simple calculations. Partial least squares with least angle regression–least absolute shrinkage and selection operator wavelength selection were used to determine melamine with increased accuracy. The limits of qualitative and quantitative detection were 25 ppm. These methods may be used to analyze other adulterated feed with low concentrations of contaminants.
Acknowledgment
The authors are grateful to the Chinese Academy of Agricultural Sciences for samples collection and PerkinElmer for technical support.