Study On Fingerprint Of Chinese Traditional Medicine Using High Performance Liquid Chromatography With The Aid Of Chemometrics

With the rapid development of hyphenated instruments and related chemometric methods,it is possible to provide powerful tools and extensive prospect for the building and evaluating ofTraditional Chinese Medicine (TCM) fingerprint. In this paper, with the help of HPLC-DADseparation and detection combining related chemometric methods, the chromatographicfingerprints of Huoxiang Zhengqi Tincture, Sanhuang tablets and Qing-Re-Jie-Du Oral Liquidhave been built. Then, they were evaluated by the chemometric methods. It includes fourchapters as follows:In Chapter one, a review on research status and development foreground was arised, whichthe conception, characteristics, research content and its importance for quality control of TCMwere introdced mainly. In addition, the methods applied for fingerprint analysis of TCM werealso reviwed, and the chemometrics methods were discussed specially. The trend of fingerprintanalysis was mentioned in the end.In-Chapter two, several chemometrics techniques were emloyed to investigate thechromatographic fingerprint data of Huoxiang Zhengqi Tincture (HZT) in order to compare thedifferences and similarities among the 21 samples from different manufacturers and batches.The results showed that similarity analysis, hierarchical cluster analysis (HCA) andmulti-criteria decision making methodology (MCDM), PROMETHEE and GAIA, weresuccessfully used to discriminate HZT samples from three different manufacturers. Furthermore, Fourier Transform-Infrared (FT-IR) Spectra were established for 3 of 21 samples, andtheir FT-IR fingerprints were compared by the common similarity method with the HPLCresults. It was found that the FT-IR spectra did not discriminate the samples from the differentmanufacturers and FT-IR was not suitable to develop the fingerprint of HZT samples. Themethod developed in this paper could be effectively used to control the quality of HuoxiangZhengqi Tincture, providing a reference method for chemical pattern recognition of other TCMproducts.In chapter three, the high performance liquid chromatographic fingerprint of Sanhuangtablets was developed and t-testing method was proposed to evaluate the similarities anddifferences among samples, which was possible to distinguish the fingerprints. Then, thetechnique of projection discriminance based on principal component analysis (PCA) wasapplied to verify the results obtaind with t-testing method. Further more, chemical patternrecognition of the fingerprints was carried out using k-means Clustering method with the number of cluster obtained by t-testing method and PCA. The results showed that t-testingmethod was capable of evaluating the similiarities and differences of TCM, and the clusterresults obtaind by k-means clustering method were similar to which obtained by PCA.In chapter four, three typical neural networks, including Back Propagation Network (BPN),Radial Basis Network (RBN) and Self-Organized Competition Network (SOCN), were emplyedto deal with the high performance liquid, chromatographic fingerprint data of Qing-Re-Jie-DuOral Liquid, and leave-one-out method was applied to evaluate their ability of studing andgeneralizing. It was found that Self-organized competition network not only has strong abilityof self-studing which provding an reasonable clustering result, but also stronger ability ofgeneralizing than other two kinds of networks which forecasting unknown samples moreaccurately.