| Chinese medicinal materials are a widely-used natural resource for disease treatment.Compared to Western medicine,they have the advantage of lower toxicity and fewer side effects.However,due to their complex composition,studying them poses a challenge.Accurate detection of their components is crucial for ensuring the quality and effectiveness of traditional Chinese medicine treatment.There are some methods available for quantitative analysis of target components,including liquid chromatography,infrared spectroscopy,fluorescence spectroscopy,and Raman spectroscopy.Among these,fluorescence spectroscopy is increasingly being used for its advantages of high sensitivity,fast detection,low cost,and environmental friendliness.This thesis aims to develop chemometric methods and apply them to rapid and accurate quantitative analysis of several effective components in different Chinese medicine systems using fluorescence spectroscopy.This research has both theoretical and practical significance,mainly involve the following aspects:Chapter 1 first elaborates on the importance of quality control in Chinese medicine and emphasizes the practical significance of fast detection of its components.Then,it briefly introduces various spectroscopic techniques and specifically highlights the application of fluorescence spectroscopy in the analysis of Chinese medicine components.Finally,it provides a detailed introduction to the analysis methods that applied in fluorescence spectroscopy data.Chapter 2 uses different chemometrics methods,including Lasso regression(LAR),Interval Partial Least Squares(i PLS),and multidimensional partial least squares(N-PLS),along with flourescence excitation-emission matrix(EEM)spectroscopy,to accurately quantify aesculin and aesculetin in Chinese medicine Cortex fraxini.Despite the strong overlap of EEM spectroscopy of the two analytes,satisfactory quantitative analysis results were obtained.This study proves that the N-PLS algorithm combined with EEM can obtain the most accurate quantitative analysis results among the three algorithms.Moreover,the effectiveness of the proposed methods were confirmed by using the ultra-high performance liquid chromatography(UHPLC)to analyze the same samples.Chapter 3,a new method was developed for analyzing target compounds in Chinese medicine Mangnolia officinalis by combining a newly constructed EEM spectra with N-PLS and Partial Least Squares(PLS)methods.The new EEM spectra was created using a novel approach to extract EEM feature spectra.The effectiveness of the method was verified using a public available dataset.The results demonstrated that the new EEM spectra yielded the same accurate analysis results as the original EEM,but with less useless information and improved calculation speed.This proposed method shows promising applications for analyzing EEM data with signal overlap.Chapter 4,due to the presence of information irrelevant to the analysis target in spectral data,inspired by the characteristic spectral extraction method in Chapter 3,this chapter proposes a new EEM characteristic spectral extraction method and uses a strategy of algorithm combination to establish a quantitative model.Firstly,the LAR,Ridge Regression(RR),and Multiple Linear Regression(MLR)methods are used to directly construct a quantitative model,with the input being the newly extracted EEM spectrum.Secondly,the i PLS and Stepwise Regression methods are used to extract spectral features,and a quantitative model is built by combining the LAR,RR,and MLR methods.By comparing these two methods,the most suitable data analysis method for each system is selected,realizing fast and accurate quantitative analysis,and providing new ideas for solving signal overlap problems in spectral data. |
PDF Full Text Request