Hits:

Indexed by:期刊论文

Date of Publication:2012-02-05

Journal:JOURNAL OF PHARMACEUTICAL AND BIOMEDICAL ANALYSIS

Included Journals:SCIE、PubMed、Scopus

Volume:59

Issue:1

Page Number:44-49

ISSN No.:0731-7085

Key Words:Near-infrared spectroscopy classification; Quantification; Corydalis Tuber; Partial least squares regression

Abstract:With the application of near-infrared spectroscopy (NIRS), a convenient and rapid method for determination of alkaloids in Corydalis Tuber extract and classification for samples from different locations have been developed. Five different samples were collected according to their geographical origin, 2-Der with smoothing point of 17 was applied as the spectral pre-treatment, and the 1st to scaling range algorithm was adjusted to be optimal approach, classification model was constructed over the wavelength range of 4582-4270 cm(-1), 5562-4976 cm(-1) and 7000-7467 cm(-1) with a great recognition rate. For prediction model, partial least squares (PLS) algorithm was utilized referring to HPLC-UV reference method, the optimum models were obtained after adjustment. Pre-processing methods of calibration models were COE for protopine and min-max normalization for palmatine and MSC for tetrahydropalmatine, respectively. The root mean square errors of cross-validation (RMSECV) for protopine, palmatine, tetrahydropalmatine were 0.884, 1.83, 3.23 mg/g. The correlation coefficients (R(2)) were 99.75, 98.41 and 97.34%. T test was applied, in the model of tetrahydropalmatine; there is no significant difference between NIR prediction and HPLC reference method at 95% confidence interval with t = 0.746 < t((0,05.20)) = 2.086, therefore NIRS is a reliable analytical tool in establishing prediction models. (C) 2011 Elsevier B.V. All rights reserved.