Publications serve as the concrete art form for the scientist. It is his modus operandi. Authorship is akin to success and achievement. It cannot and should not deteriorate into a bargaining tool or commodity.
Tankeu, S., Vermaak, I., Chen, W., Sandasi, M., Viljoen, A.M. 2016. Application of hyperspectral imaging for the differentiation between two “fang ji” herbal medicines: Stephania tetrandra and the nephrotoxic Aristolochia fangchi. Phytochemistry 122: 213–222.
The quality control of medicinal plants remains challenging due to the complexity of the botanical matrix. In addition, the innate variability in phytochemical profiles and toxicity issues associated with species substitution and adulteration necessitate the development of reliable methods for the quality assessment of herbal medicines. The analytical methods commonly used to determine the quality of plant materials such as liquid chromatography coupled to mass spectrometry (LC-MS) are expensive, laborious and require skilled analysts. In this study, hyperspectral imaging in combination with a chemometric modeling tool, partial least square discriminant analysis (PLS-DA), is suggested as an alternative, inexpensive and simple method to distinguish between Stephania tetrandra and nephrotoxic Aristolochia fangchi root powder. Stephania tetrandra ("hang fang ji”) is a traditional Chinese medicine which functions as a diuretic and antirheumatic medicine. It is mistakenly substituted and adulterated with the nephrotoxic A. fangchi ("guang fang ji”) which may cause severe health consequences. Short wave infrared (SWIR) hyperspectral images of powdered S. tetrandra and A. fangchi were obtained in the wavelength region of 920 – 2514 nm. The range containing the discriminating information was identified as 964 – 1774 nm. After reduction of the dimensionality of the data based on the selected discrimination information range, a discrimination model was created with a correlation of determination (R2) = 0.9 and a root mean square error of prediction (RMSEP) of 0.23 for the two species. The constructed model was successfully used in the identification of A. fangchi and S. tetrandra samples inserted into the model as an external validation set. In addition, adulteration detection was investigated by preparing incremental adulteration mixtures of S. tetrandra with A. fangchi (10 to 90%). Based on pixel abundance, hyperspectral imaging showed the ability to accurately predict adulteration as low as 10%. It is evident that hyperspectral imaging has tremendous potential in the development of visual, easy-to-apply quality control methods.