Department of Natural Medicine, West China School of Pharmacy, Sichuan University, and Key Laboratory of Drug Targeting, Ministry of Education, No. 17, Section 3, Ren-Min-Nan-Lu Road, Chengdu, Sichuan 610041, PR China.
Department of Natural Medicine, West China School of Pharmacy, Sichuan University, and Key Laboratory of Drug Targeting, Ministry of Education, No. 17, Section 3, Ren-Min-Nan-Lu Road, Chengdu, Sichuan 610041, PR China.
Spectrochim Acta A Mol Biomol Spectrosc. 2018 Nov 5;204:131-140. doi: 10.1016/j.saa.2018.06.004. Epub 2018 Jun 2.
Rhodiola is an increasingly widely used traditional Tibetan medicine and traditional Chinese medicine in China. The composition profiles of bioactive compounds are somewhat jagged according to different species, which makes it crucial to identify authentic Rhodiola species accurately so as to ensure clinical application of Rhodiola. In this paper, a nondestructive, rapid, and efficient method in classification of Rhodiola was developed by Fourier transform near-infrared (FT-NIR) spectroscopy combined with chemometrics analysis. A total of 160 batches of raw spectra were obtained from four different species of Rhodiola by FT-NIR, such as Rhodiola crenulata, Rhodiola fastigiata, Rhodiola kirilowii, and Rhodiola brevipetiolata. After excluding the outliers, different performances of 3 sample dividing methods, 12 spectral preprocessing methods, 2 wavelength selection methods, and 2 modeling evaluation methods were compared. The results indicated that this combination was superior than others in the authenticity identification analysis, which was FT-NIR combined with sample set partitioning based on joint x-y distances (SPXY), standard normal variate transformation (SNV) + Norris-Williams (NW) + 2nd derivative, competitive adaptive reweighted sampling (CARS), and kernel extreme learning machine (KELM). The accuracy (ACCU), sensitivity (SENS), and specificity (SPEC) of the optimal model were all 1, which showed that this combination of FT-NIR and chemometrics methods had the optimal authenticity identification performance. The classification performance of the partial least squares discriminant analysis (PLS-DA) model was slightly lower than KELM model, and PLS-DA model results were ACCU = 0.97, SENS = 0.93, and SPEC = 0.98, respectively. It can be concluded that FT-NIR combined with chemometrics analysis has great potential in authenticity identification and classification of Rhodiola, which can provide a valuable reference for the safety and effectiveness of clinical application of Rhodiola.
红景天是一种在中国被广泛应用的传统藏药和中药。由于不同物种的生物活性化合物组成谱有些参差不齐，因此准确识别正品红景天物种至关重要，以确保红景天的临床应用安全有效。在这项研究中，我们建立了一种无损、快速、高效的红景天分类方法，即傅里叶变换近红外（FT-NIR）光谱结合化学计量学分析。我们共获得了 160 批来自四个不同物种（大花红景天、唐古特大红景天、高山红景天和长鞭红景天）的红景天原始光谱。在剔除异常值后，比较了 3 种样品分割方法、12 种光谱预处理方法、2 种波长选择方法和 2 种建模评价方法的不同性能。结果表明，FT-NIR 与基于联合 x-y 距离的样品集分割（SPXY）、标准正态变量变换（SNV）+Norris-Williams（NW）+二阶导数、竞争自适应重加权采样（CARS）和核极端学习机（KELM）相结合的方法在真实性识别分析中表现更优。最佳模型的准确率（ACCU）、灵敏度（SENS）和特异性（SPEC）均为 1，表明 FT-NIR 与化学计量学方法的结合具有最佳的真实性识别性能。偏最小二乘判别分析（PLS-DA）模型的分类性能略低于 KELM 模型，PLS-DA 模型的 ACCU 为 0.97、SENS 为 0.93、SPEC 为 0.98。综上所述，FT-NIR 结合化学计量学分析在红景天的真实性识别和分类方面具有巨大潜力，可为红景天的临床应用的安全性和有效性提供有价值的参考。
