Discrimination of traditional herbal medicines based on terahertz spectroscopy

Terahertz (THz) spectroscopy was employed to develop an efficient and applicative way of discriminating traditional herbal medicines in this paper. Spectra of three different herbal medicines (Herba Solani Lyrati, Herba Solani Nigri and Herba Aristolochiae Mollissimae) were obtained in the range of 0.2-1.2 THz. Principal component analysis (PCA) was applied to reduce the dimensionality of original spectral information. Three classification algorithms, support vector machine (SVM), decision tree (DT), and random forest (RF) were used to discriminate the herbal medicines. Receiver operating characteristic (ROC) curve and area under the ROC curve (AUC) were combined with classification accuracy to evaluate the performances of the three classification algorithms. The PCA-RF method got the best ROC curve and AUC, and achieved a prediction accuracy of 99%. The experimental results indicate that THz spectroscopy combined with chemometric algorithms is an effective and rapid method for the discrimination of traditional herbal medicines.