Surface-enhanced Raman spectroscopy for classification of testosterone propionate and nandrolone residues in chicken

Abstract There is a critical need for a rapid and simple screening method of androgens in chicken. In this study, we evaluated surface-enhanced Raman spectroscopy (SERS) coupled with multivariate techniques for the classification of two androgens (i.e., testosterone propionate and nandrolone) in chicken from 294 samples. Raw Raman spectra were pretreated by using the methods of baseline correction, normalization and second derivative. Support vector machines (SVM) model using the score values of the first four principal components as the inputs was developed to classify all the chicken samples into the four groups (i.e., control, nandrolone, testosterone propionate, and testosterone propionate combined with nandrolone groups) with accuracy of 96.9%. Furthermore, the particle swarm optimization (PSO) was adopted to automatically optimize the penalty parameter C and the kernel parameter g of SVM model for improving the classification accuracy. The experimental results demonstrated that SERS, in combination with multivariate methods, could be utilized as a rapid and simple classification assay of androgens in chicken and exhibited great potential in practical applications as a screening tool to better serve customers.