Application Of Near Infrared Spectroscopy In Analysis Of Cordyceps Militaris Mycelium

Cordyceps is rare and precious wild Chinese herb, which has the effect of improving deficiency syndromes, protecting lungs, invigorating kidneys, stopping bleeding, dissipating phlegm, nourishing the body, reinforcing the body resistance, etc. It ranks among the three major treasured tonics, together with ginseng and deer antler. Cordyceps Militaris is also receiving increasing attention in the world, which provides a new source of healthcare products and medicines. In Cordyceps Militaris, adenosine, polysaccharides, Cordycepic acid and protein are four major active ingredients whose contents have a direct effect on its quality and pharmacological activity. When it comes to measure their contents, traditional chemical analysis methods are time consuming, costly, destructive and have poor repeatable accuracy. Still, they tend to cause pollution to environment. Therefore, it is necessary to explore a rapid and undamaged analysis method to measure effective ingredients of Cordyceps Militaris. NIRS is a new technique that can measure chemical ingredients of substance non-destructively and rapidly by making a suitable model using optical properties of substance in the near infrared region, which can meet the new requirements for drug analysis.The thesis started with the research on original strains of Cordyceps Militaris. Through chemical mutagenesis, a large number of mutagenesis Cordyceps Militaris strains were got out of the original strains. After they were cultured in shake-flask, the accumulation of Cordyceps Militaris samples for NIRS analysis was completed. Then the contents of adenosine, polysaccharides, Cordycepic acid and protein in different batches of Cordyceps Militaris mycelia were measured with HPLC, colorimetric method, anthrone-sulfuric acid method and Kjeldahl method.With near-infrared spectroscopy analysis, the near-infrared spectra of 389 different samples of Cordyceps Militaris mycelia were got. Then modern metrology methods were used to make PLS and RBFNN models for the quantitative analysis of the contents of adenosine, polysaccharides, Cordycepic acid and protein. After assessing the models with objective function and comparative analysis, the optimal models were determined. In the process of establishing the PLS quantitative analysis model, principal ingredients were first extracted from the original spectra and different pre-processing spectra. Subsequently, the effects of the number of major factors on the models were studied. When the RBFNN quantitative analysis models were developed, principal ingredients were also extracted from the original spectra and different pre-processing spectra, based on which the influence of input nodes, hidden nodes and spread constant on the models was examined. The experimental results showed that the former 15 principal component scores could take the place of the original spectral information effectively. As for the research on near-infrared spectra, some common pretreatment methods such as Savitzky-Golay smoothing, Fast Fourier transform, derivative, Wavelet transform were conducted. The results showed that Savitzky-Golay smoothing, Fast Fourier transform and Wavelet transform were effective in eliminating the noise from spectrum and derivative method could remove the interference from the baseline and other backgrounds effectively.The research results of the thesis are the following:(1) Effects of pretreatment methods on PLS and RBFNN quantitative analysis models were studied. The results showed that after eliminating the interference from random noise, baseline and other backgrounds, the stability, fit and predictive capability of the models could be improved significantly.(2) Effects of parameters on PLS and RBFNN quantitative analysis models were studied. The results showed that quantitative analysis models built on the optimal parameters improved significantly.(3) The optimal models to measure the contents of four ingredients in Cordyceps Militaris mycelia were determined.The optimal quantitative analysis model to measure the content of adenosine in Cordyceps Militaris mycelia was as follows: With the original spectra the PLS quantitative analysis model was established. The best factor was 10. RMSECV, Rv, RMSEP and Rp were 0.6520, 0.9014, 0.5780 and 0.9297 respectively.The optimal quantitative analysis model to measure the content of protein in Cordyceps Militaris mycelia was as follows: With first derivative spectra (window size 20) the RBFNN quantitative analysis model was established. The best input nodes, hidden nodes and spread constant were 8, 20 and 2.5. RMSECV, Rv, RMSEP and Rp were 0.0218, 0.9801, 0.0202 and 0.9785 respectively.The optimal quantitative analysis model to measure the content of Cordycepic acid in Cordyceps Militaris mycelia was as follows: With wavelet transform spectra (4 scale under db2 function) the PLS quantitative analysis model was built. The best factor of the model was 10. RMSECV, Rv, RMSEP and Rp were 0.0088, 0.8807, 0.0087 and 0.9147 respectively.The optimal quantitative analysis model to measure the content of polysaccharides in Cordyceps Militaris mycelia was as follows: With the original spectra the PLS quantitative analysis model was established. The best factor was 10. RMSECV, Rv, RMSEP and Rp were 0.0091, 0.8871, 0.0079 and 0.9261 respectively.The research results show that NIRS is effective in measuring the contents of adenosine, polysaccharides, Cordycepic acid and protein in Cordyceps Militaris mycelium under suitable conditions, which is of higher applicable value for the non-destructive and rapid measurement of Medicinal fungi.