Rapid and undamaged identification of the Semen cuscutae and its adulterants based on image analysis and electronic nose analysis

Abstract

In this paper, image analysis and electronic nose analysis were used to develop a rapid, reliable and undamaged method of identifying the Semen cuscutae and its adulterants including radish seed and Sinapis alba seeds. The results showed that the highest identification rate was 100% for the training set and 96.5% for the test set based on image analysis and various chemometric techniques including principal component analysis, linear discriminant analysis (LDA), k-nearest neighbor, random forests, artificial neural network and support vector machine analysis. LDA analysis based on electronic nose analysis exhibited better discrimination result, ranging from 95.5–100% for the correct classification rate and 95.4–100% for the cross-validation rate, respectively. LDA model based on electronic nose data from 16 to 30 s was the best, and both the correct classification rate and cross-validation rate reached 100%. These results provided a simple, fast and non-destructive method to identify the true and false of Semen cuscutae, which can serve as a reference to identify the authenticity of the medicinal plants.