Influence Factors Of The Near Infrared Spectra Of Traditional Chinese Medicine Research

Near Infrared Spectroscopy (NIRS) is a fast, non-destructive and green analysis technique. A complete near-infrared spectroscopy method includes data collection, model calibration and validation, following model maintenance and updating. For any analysis method, the accuracy of the raw data is the guarantee to ensure the accuracy and reliability of the analytical methods. Though chemometrics can be utilized to greatly improve the predictability of the NIR model, the reliability of the raw data is the most important. There are many factors affecting the spectral response, and for specific instruments and analysis environment, sample status and loading condition are the most important influencing factors. This paper is to explore the effects of sample presentation, including the optical path of the liquid sample and particle size of the powder sample on near infrared transmission spectroscopy and near-infrared diffuse reflectance spectroscopy, respectively, in order to ensure a strong spectral absorbance and accuracy of near-infrared spectroscopy in quality control of traditional Chinese medicine. The mainly research contents are as follows:I. The influence of the optical path length on near infrared transmission spectroscopyNIR transmission spectra of Qingkailing injection were collected with different optical path lengths (6and8mm) using Fourier transform near infrared (FT-NIR) spectrometer in the wavelength range of4000cm-1to10000cm-1. Calibration models were developed using partial least-squares regression (PLSR) with high-performance liquid chromatography (HPLC) as a reference method. Results showed that the PLS model of raw spectra with6mm optical path length presented the best prediction ability. For the calibration set, Re=0.989, RMSEC=0.16, and for the validation set, Rp=0.993, RMSEP=0.12, RPD=7.65. It was illustrated that optical path length affected the NIR analysis result of baicalin determination in Qingkailing injection, and appropriate optical path length could guarantee the detection accuracy.Ⅱ. Effects of granularity of the samples on the near-infrared diffuse reflectance spectroscopyFirstly, spectra reproducibility after sample repackage was examined. When sample particle size is large, the spectral reproducibility was poor and related to different bands. When the particle size was small, the reproducibility tend to be good in all spectral bands.Secondly, a variety of Chinese herbal medicines were used to study the effect of particle size on the absorbance of different near-infrared spectral bands. The results showed that:(1) Effects of particle size on the near-infrared spectroscopy were different according to different bands(combination region, first combination-overtone region and second combination-overtone region). In the CR region and FCOR region, spectroscopy intensity is proportional to the particle size and influence of particle size was greater as the wavelength increased; while in SCOR region,the effects were various for different Chinese herbal medicine.(2) For different types of herbs, particle size effects variate greatly according to the physical structure.Lastly, the effect of sample particle size on the measurement of SSA in Bnpleurum chimnse DC by near-infrared reflectance (NIR) spectroscopy was explored. Four type samples of different granularity were prepared including powder samples passed through40mesh,65meshu80mesh and100mesh sieve. Single PLS model was constructed separately for every kind of sample, and data preprocessing techniques were supplemented to optimize calibration model. The65-mesh model exhibit the best prediction ability with RMSEP=0.492, Rp=0.9221and RPD=2.58. Furthermore, a granularity-hybrid calibration model was developed by incorporating granularity variation, and it showed better performance than single model. And still the65-mesh samples were predicted the most accurately with RMSEP=0.481, RP=0.9279and RPD=2.64. All the above results present guidance for sample preparation in NIR analysis of Chinese traditional medicine and a reference for construction of a robust model eliminating physical factors.Ⅲ. Quality evaluation of traditional Chinese medicine based on near-infrared spectra of the original sample surface(1) Discrimination of genuine Chinese Yam with NIR spectroscopy. Ninety samples from three origins were collected and their near infrared diffuse reflectance spectra were acquired. Discriminant analysis was employed to create qualitative model, and several data pre-processing methods were explored. The results showed that genuine Chinese Yam was greatly different from other samples according to score plot of PCA. The discriminant analysis model with raw spectra after pretreatment of multiplicative scatter correction,2nd derivative and Savitzky-Golay filter smoothing, showed the best predicative ability (identification rate were both100%for the calibration sample set and validation sample set).(2) Identification of the sulfur fumigated Radix Puerariae based on near-infrared spectroscopy. Thirty samples were randomly selected from each kind of sample. And near infrared spectra of the cross-sectional and longitudinal section of sample were acquied. Principal component analysis was conducted, and it was observed from the score plot that sulfur fumigated Radix Puerariae and the not sulfur fumigated ones were obviously divided into two categories.50discriminant modelswere established for every type of spectra and box plot was used to conductdescriptive analysis. For the50models constructed with cross-section, longitudinal spectra and total spectra, the recognition rate were94.4±1.33%, 95.6±1.50%, and95.3±1.29%, respectively. And further the non-parametric test (Mann-Whitney U test) was applied to perform difference analysis. The results showed that the P=0.256>0.05and no difference was observed with α=0.05.(3) Rapid identification of Wolfberry Fruit of different geographic regions with sample surface near infrared spectra combined with multi-class SVM. Portable near infrared spectrometer combined with multi-class support vector machines was used to discriminate Wolfberry Fruit of different geographic regions. To eliminate the influence of sample subset partitioning on model performance, multiple modeling and predicting were conducted and the statistical result of identification rate was utilized to assess model performance of different acquisition sites. The results showed that SVM model with raw spectra after pretreatment of second derivative and Savitzky-Golay filter smoothing, showed the best predicative ability. And model of every acquisition site except the site5exhibit good stability and prediction ability and its median and average of identification rate of external validation were all greater than97%. It was suggested that surface NIR spectra of Wolfberry Fruit was applicable to accurately identification of geographic region, and portable near infrared spectrometercould act as an effective means of monitoring the quality of Chinese herbal medicine in circulation.