Application of terahertz spectroscopy imaging for discrimination of transgenic rice seeds with chemometrics

Discrimination of genetically modified organisms is increasingly demanded by legislation and consumers worldwide. The feasibility of a non-destructive discrimination of transgenic rice seeds from its non-transgenic counterparts was examined by terahertz spectroscopy imaging system combined with chemometrics. Principal component analysis (PCA), least squares support vector machines (LS-SVM), PCA-back propagation neural network (PCA-BPNN), and random forest (RF) models with the first and second derivative and standard normal variate transformation (SNV) pre-treatments were applied to classify rice seeds based on genotype. The results demonstrated that differences between non-transgenic and transgenic rice seeds did exist, and an excellent classification (accuracy was 96.67% in the prediction set) could be achieved using the RF model combined with the first derivative pre-treatment. The results indicated that THz spectroscopy imaging together with chemometrics would be a promising technique to identify transgenic rice seeds with high efficiency and without any sample preparation.