Wang Yuan, Qin Min-Jian, Qi Jin, Yu Bo-Yang, Tang Li
College of Traditional Chinese Medicine, China Pharmaceutical University, Nanjing 210009, China.
Guang Pu Xue Yu Guang Pu Fen Xi. 2009 Oct;29(10):2677-80.
In the present paper the polysaccharides in Ophiopogon japonicus was analysed quantitatively by using near infrared spectroscopy. The spectral characteristics of the primary ingredients in ophiopogonis were obtained by applying first derivative, second derivative, smoothing, SNV (standard normal variate), and MSC (multiple scatter correction), and the wavelengths were selected for the best model. Meantime, in combination with the PLS algorithm the calibration process was performed for the quantitative analysis of the polysaccharides in Ophiopogon japonicus. The result showed that after optimizing all the factors, the best model of equation was using "first derivative" + "MSC" + "SG", and the wavelengths for the best model were selected in 4,000-4,900, 5,100-6,900 and 7,050-10,000 cm(-1). The result showed a fine precision of the method, with R2, RMSEC, R2CV, RMSECV and principal factors being 0.996, 0.237, 0.973, 0.583 and 6, respectively. A set of representative samples were used to check the model, and the prediction coefficient of determination was 96.88%.
本文采用近红外光谱法对麦冬中的多糖进行了定量分析。通过应用一阶导数、二阶导数、平滑处理、标准正态变量变换（SNV）和多元散射校正（MSC）获得了麦冬主要成分的光谱特征，并选择了最佳模型的波长。同时，结合偏最小二乘法（PLS）算法对麦冬多糖进行定量分析的校准过程。结果表明，在对所有因素进行优化后，最佳的方程模型为采用“一阶导数”+“MSC”+“Savitzky-Golay平滑（SG）”，最佳模型的波长选择在4000 - 4900、5100 - 6900和7050 - 10000 cm⁻¹。结果表明该方法具有良好的精密度，决定系数R²、校正均方根误差（RMSEC）、交叉验证决定系数（R²CV）、交叉验证均方根误差（RMSECV）和主因子分别为0.996、0.237、0.973、0.583和6。使用一组代表性样品对模型进行检验，预测决定系数为96.88%。
