Rapid identification and quantification of sesame oils adulteration using low frequency dielectric spectroscopy combined with chemometrics

Adulteration is a serious problem in the expensive edible oils, especially for sesame oil (SEO). SEO is widely adulterated with cheap edible oils such as rapeseed oil (RO). Therefore, developing a low-cost, practical, and rapid analytical method for detecting and quantifying such adulteration in SEO would be useful. A dielectric spectroscopy-based system in the range of 40 kHz-20 MHz was developed to detect SEO adulteration. The principal component analysis (PCA) model was developed for detecting the authentic SEO. Amount of adulteration was predicted by means of artificial neural network (ANN) and support vector regression (SVR). Out-of-sample validation indicated the radial basis function (RBF) kernel SVR model had the best performance with correlation coefficient (R) of 0.9604, mean absolute error (MAE) of 4.9888 (mL/100 mL) and Root Mean Square Error (RMSE) of 6.0129 (mL/100 mL) followed by the ANN model with R value of 0.9192, MAE of 6.4033 (mL/100 mL) and RMSE of 7.8072 (mL/100 mL). The results showed the developed technique as a simple and inexpensive tool has an acceptable capability for detecting RO impurities in SEO. However, further research is needed to find whether the developed system would be able to detect adulterations generated by other inexpensive edible oils in SEO.

Automated summary

A new method was developed for detecting adulteration in sesame oil, using PCA model for authentic sesame oil detection and predicting adulteration amounts through artificial neural network and support vector regression. The radial basis function kernel SVR model had the best performance, followed by the ANN model.