温度限制串联相关网络-近红外光谱法用于药物甲硝唑的质量控制

近红外漫反射光谱是一种简便、快速的有机物分析方法,样品不需处理即可直接测量,易于实现固态样非破坏测定.近红外漫反射光谱分析技术广泛应用于农业、食品、化妆品、烟草和石油等方面的组分分用近红外漫反射光谱法进行药品的非破坏性分析正成为国际热门课题.但近红外漫反射光谱的光谱范宽,吸收强度很弱,且组分间光谱严重重叠,给非破坏性分析带来了困难.而近红外漫反射光谱法与化量学相结合,能有效地解决光谱重叠带来的问题[1～3].     

Identification of rhubarbs by using NIR spectrometry and temperature-constrained cascade correlation networks

Temperature-constrained cascade correlation networks (TCCCNs) were used to identify powdered rhubarbs based on their near-infrared spectra. Different network configurations that used multiple network models with single output (Uni-TCCCN) and single networks with multiple outputs (Multi-TCCCN) were compared. Comparative studies were made by using Latin-partitions and leave-one-out cross-validation methods. Results showed that multiple networks with single output predicted generally better than single network with multiple outputs. Better results with TCCCN models were obtained compared with conventional back propagation neural networks (BPNNs). The effects of parameters on correct identification and parameter optimizations were discussed in detail. With optimized neural network training parameters, NIR spectra from powdered rhubarb samples were classified by a TCCCN model with 100% accuracy.     

中药口服固体制剂原辅料近红外光谱数据库的构建及应用

建立了中药口服固体制剂原辅料近红外(NIR)光谱数据库,采用模式识别方法研究了NIR光谱数据在物料分类和物性预测中的应用。使用便携式近红外光谱仪快速测量149批原辅料粉末的NIR漫反射光谱数据,并录入iTCM数据库。利用主成分分析(PCA)法探究NIR光谱数据对已知结构物料的分类能力,采用偏最小二乘(PLS)法研究了NIR光谱对原辅料物性参数和直接压片片剂性能的预测能力。经标准正态变量变换(SNV)+Savitzky-Golay(SG)平滑+一阶导数处理后的NIR光谱数据对微晶纤维素、乳糖、乙基纤维素、交联聚维酮和羟丙基甲基纤维素这5类辅料的区分能力较好。NIR光谱数据与原辅料粉末粒径、密度和吸湿性的相关性较强。NIR光谱信息作为物料物理性质的补充,可提高粉末直接压片片剂性能预测模型的性能。NIR光谱数据是iTCM数据库物性参数数据的补充,物性参数与NIR光谱数据的结合能更全面地表征原辅料的性质。     

Influence of the procedure used to prepare the calibration sample set on the performance of near infrared spectroscopy in quantitative pharmaceutical analyses

Calibrating near infrared diffuse reflectance spectroscopy (NIRS) methods usually involves preparing a set of samples with a view to expanding the analyte concentration range spanned by production samples. In this work, the performances of the two procedures most frequently used for this purpose in near infrared pharmaceutical analysis, viz., synthetic samples obtained by weighing of the pure constituents of the pharmaceutical and doped samples made by under- or overdosing previously powdered production samples, were compared. Both procedures were found to provide similar results in the quantification of the active compound in the pharmaceutical, which was determined with a relative standard error of prediction (RSEP) of < 1.6%. However, the two types of sample preparation provide different spectra, which precludes the accurate quantification of synthetic samples from calibrations obtained with doped samples and vice versa. None of the mathematical pre-treatments tested with a view to reducing this different scattering (viz., second derivative, standard normal variate and orthogonal signal correction) could effectively solve this problem. This hinders accurate validation of the linearity of the procedure and makes it advisable to use doped samples which are markedly less different to production samples.     

Validation using sensitivity and target transform factor analyses of neural network models for classifying bacteria from mass spectra

Temperature constrained cascade correlation networks (TCCCNs) are computational neural networks that configure their own architecture, train rapidly, and give reproducible prediction results. TCCCN classification models were built using the Latin-partition method for five classes of pathogenic bacteria. Neural networks are problematic in that the relationships among the inputs (i.e., mass spectra) and the outputs (i.e., the bacterial identities) are not apparent. In this study, neural network models were constructed that successfully classified the targeted bacteria and the classification model was validated using sensitivity and target transformation factor analysis (TTFA). Without validation of the classification model, it is impossible to ascertain whether the bacteria are classified by peaks in the mass spectrum that have no causal relationships with the bacteria, but instead randomly correlate with the bacterial classes. Multiple single output network models did not offer any benefits when compared to single network models that had multiple outputs. A multiple output TCCCN model achieved classification accuracies of 96 +/- 2% and exhibited improved performance over multiple single output TCCCN models. Chemical ionization mass spectra were obtained from in situ thermal hydrolysis methylation of freeze-dried bacteria. Mass spectral peaks that pertain to the neural network classification model of the pathogenic bacterial classes were obtained by sensitivity analysis. A significant number of mass spectral peaks that had high sensitivity corresponded to known biomarkers, which is the first time that the significant peaks used by a neural network model to classify mass spectra have been divulged. Furthermore, TTFA furnishes a useful visual target as to which peaks in the mass spectrum correlate with the bacterial identities.     

近红外漫反射一阶导数光谱法作安体舒通质量控制研究

研究了近红外漫反射一阶导数光谱法作安休舒通粉末药品质量控制的可能性，用多变量统计分类技术，从安体舒通粉末药品的近红外漫反射一阶导数光谱，成功地鉴别了真，劣和假药，结果令人满意。     

粉末药品安体舒通的无损定量分析——人工神经网络－近红外光谱法的应用

研究了应用人工神经网络进行粉末药品的无损定量分析,使用安体舒通粉末药品的近红外漫反射光谱数据建立人工神经网络模型,预测未知样品,讨论了影响网络的各参数,使用了逼近度作为网络新的评价标准。     

近红外光谱-人工神经网络法用于粉末药品阿斯匹林的非破坏定量分析

研究了应用人工神经网络进行粉末药品的非破坏定量分析。使用阿斯匹林粉末药品的近红外漫反射一阶导数光谱数据建立人工神经网络模型,预测未知样品。讨论了影响网络的各参数,使用了新的网络评价标准－逼近度。     

Discrimination of Ganoderma lucidum according to geographical origin with near infrared diffuse reflectance spectroscopy and pattern recognition techniques

A rapid and nondestructive near infrared (NIR) method combined with chemometrics was used to discriminate Ganoderma lucidum according to cultivation area. Raw, first, and second derivative NIR spectra were compared to develop a robust classification rule. The chemical properties of G. lucidum samples were also investigated to find out the difference between samples from six varied origins. It could be found that the amount of polysaccharides and triterpenoid saponins in G. lucidum samples was considerably different based on cultivation area. These differences make NIR spectroscopic method viable. Principal component analysis (PCA), discriminant partial least-squares (DPLS) and discriminant analysis (DA) were applied to classify the geographical origins of those samples. The results showed that excellent classification could be obtained after optimizing spectral pre-treatment. For the discriminating of samples from three different provinces, DPLS provided 100% correct classifications. Moreover, for samples from six different locations, the correct classifications of the calibration as well as the validation data set were 96.6% using the DA method after the SNV first derivative spectral pre-treatment. Overall, NIR diffuse reflectance spectroscopy using pattern recognition was shown to have significant potential as a rapid and accurate method for the identification of herbal medicines.     

人工神经网络—近红外光谱法用于甲氧苄胺嘧啶粉末药品的非？…

近红外(NIR)光谱分析技术已应用于制药、化妆品、烟草、食品、化学药品、聚合物、纺织品、油漆涂料、煤炭和石油工业等各个领域的质量监控.近年来,NIR光谱分析技术也应用于药品分析中,因该方法具有非破坏性,样品不需要复杂的预处理和分离即可直接测定.它可对药物进行定性和定量测定以及多晶、光学异构体和湿度的测定.近红外光谱法用于无损非破坏测定胶囊类以及片剂的研究已有报道[1,2].NIR光谱在使用中也有一定的局限性,主要是结构复杂,谱图重叠多,在进行定性和定量分析中需采用一定的数据处理才能获得准确可靠的分析结果.在定量分析中,…     

Application of Near Infrared Diffuse Reflectance Spectroscopy with Radial Basis Function Neural Network to Determination of Rifampincin Isoniazid and Pyrazinamide Tablets

Partial least squares(PLS),back-propagation neural network(BPNN)and radial basis function neural network(RBFNN)were respectively used for estalishing quantative analysis models with near infrared(NIR)diffuse reflectance spectra for determining the contents of rifampincin(RMP),isoniazid(INH)and pyrazinamide(PZA)in rifampicin isoniazid and pyrazinamide tablets.Savitzky-Golay smoothing,first derivative,second derivative,fast Fourier transform(FFT)and standard normal variate(SNV)transformation methods were applied to pretreating raw NIR diffuse reflectance spectra.The raw and pretreated spectra were divided into several regions,depending on the average spectrum and RSD spectrum.Principal component analysis(PCA)method was used for analyzing the raw and pretreated spectra in different regions in order to reduce the dimensions of input data.The optimum spectral regions and the models' parameters were chosen by comparing the root mean square error of cross-validation(RMSECV)values which were obtained by leave-one-out cross-validation method.The RMSECV values of the RBFNN models for determining the contents of RMP,INH and PZA were 0.00288,0.00226 and 0.00341,respectively.Using these models for predicting the contents of INH,RMP and PZA in prediction set,the RMSEP values were 0.00266,0.00227 and 0.00411,respectively.These results are better than those obtained from PLS models and BPNN models.With additional advantages of fast calculation speed and less dependence on the initial conditions,RBFNN is a suitable tool to model complex systems.     

Quality control of pharmaceuticals with NIR: From lab to process line

This work describes a general framework for assessing the active pharmaceutical ingredient (API) and excipient concentrations simultaneously in pharmaceutical dosage forms based on laboratory-scale measurements. The work explores the comprehensive development of a near infrared (NIR) analytical protocol for the quantification of the API and excipients of a pharmaceutical formulation. The samples were based on a paracetamol (API) formulation with three excipients: microcrystalline cellulose, talc, and magnesium stearate. The developed method was based on laboratory-scale samples as calibration samples and pilot-scale samples (powders and tablets) as model test samples. Both types of samples were produced according to an experimental design. The samples were measured in reflectance mode in a Fourier-transform NIR spectrometer. Additionally, a new method for determining the minimum number of calibration samples was proposed. It was concluded that the use of laboratory-scale samples to construct the calibration set is an effective way to ensure the concentration variability in the development of calibration models for industrial applications. With this method, both API and excipients can be determined in high-throughput applications in the pharmaceutical industry.     

Discrimination of Ephedra plants with diffuse reflectance FT-NIRS and multivariate analysis

The secondary metabolites of different Ephedra plants are various. Therefore, the discrimination of different Ephedra plants is significant. An objective, easy-to-use, rapid and pollution-free approach is proposed for discriminating Ephedra plants of different species, habitats and picking times on the basis of diffuse reflectance Fourier transform near infrared spectroscopy (FT-NIRS) measurements and multivariate analysis. The Fourier transform near infrared diffuse reflectance spectra (NIRDRS) were acquired from 37 pulverized samples of Ephedra plants put in glass vials in the near infrared (NIR) region between 10 000 and 4000 cm⁻¹, averaging 64 scans per spectrum at a resolution of 4 cm⁻¹. After spectra processing and data pre-processing, spectral data were analyzed respectively with three multivariate analysis techniques: discriminant analysis (DA), self-organizing map (SOM) and back-propagation artificial neural network (BP-ANN). The proposed method could distinguish not only the Ephedra plants of three species and two habitats but also the plants picked at different times of day without special sample treatment and the use of chemical reagents. The performance indexes of the DA model were 84.2-91.9% and the prediction accuracies of both the SOM and the BP-ANN models reached 93.3-100.0%.     

Evaluation of gel permeation chromatography to Fourier transform infrared spectrometric detection using a modified thermospray for analysis of polystyrene samples

A heated gas flow modified thermospray was used to couple gel permeation chromatography (GPC) to Fourier transform infrared spectrometry (FTIR) for the analysis of the standard polystyrene samples. Effluents from the GPC column were evaporated and the solutes were deposited as a series of spots on the surface of a moving stainless steel belt (0.025 mm thickness × 13 mm width). The belt continuously transferred the spots into the diffuse reflectance (DRIFT) accessory of the FTIR spectrometer, enabling identification of the deposited solutes by measurement of the diffuse reflectance IR spectrum. The IR spectra of the separated components showed excellent agreement of the spectral features to those of standard FTIR spectra and no thermal degradation was observed.     

基于近红外光谱技术与化学计量学的绿茶无损鉴别方法研究

该文利用近红外光谱技术结合化学计量学方法开发了不同品种绿茶的无损鉴别方法。通过近红外光谱技术得到了8个品种绿茶样品的近红外光谱,比较了单一以及优化组合光谱预处理方法对光谱的影响,利用无监督的主成分分析(PCA)与有监督的线性判别分析方法(LDA)分别构建了茶叶品种鉴别模型。结果表明：对比单一预处理方法,优化组合预处理具有更优的鉴别准确性。标准正态变量变换预处理消除了茶叶样品大小不均造成的光谱散射影响,一阶导数预处理实现了变动背景的消除,减少了基线漂移的影响,突出了图谱中的有效信息,采用二者相结合的预处理方式并结合无监督的主成分分析法可实现较为准确的绿茶样品种类鉴别分析,准确率达75.0%。此外,采用有监督的线性判别分析方法处理原始光谱数据,可达到100%的鉴别准确率,但该方法需提供类别的先验知识。因此,采用近红外光谱技术和化学计量学相结合的手段可实现不同品种绿茶的快速无损鉴别。     

Determination of ascorbic acid in pharmaceutical preparations by near infrared reflectance spectroscopy

We developed a method for determination of ascorbic acid in pharmaceutical preparations containing various excipients by using near infrared diffuse reflectance spectroscopy and two different calibration methods, viz. stepwise multiple linear regression (SMLR) and partial least-squares (PLS) regression, which provided comparable results and resulted in prediction errors of 1-2%. However, the PLS method provided somewhat better results with the more complex samples.     

Evaluation of gel permeation chromatography to Fourier transform infrared spectrometric detection using a modified thermospray for analysis of polystyrene samples

A heated gas flow modified thermospray was used to couple gel permeation chromatography (GPC) to Fourier transform infrared spectrometry (FTIR) for the analysis of the standard polystyrene samples. Effluents from the GPC column were evaporated and the solutes were deposited as a series of spots on the surface of a moving stainless steel belt (0.025 mm thickness × 13 mm width). The belt continuously transferred the spots into the diffuse reflectance (DRIFT) accessory of the FTIR spectrometer, enabling identification of the deposited solutes by measurement of the diffuse reflectance IR spectrum. The IR spectra of the separated components showed excellent agreement of the spectral features to those of standard FTIR spectra and no thermal degradation was observed. Received: 20 May 1996 / Revised: 17 October 1996 / Accepted: 28 November 1996     

Application of artificial neural networks to the classification of soils from São Paulo State using near-infrared spectroscopy

This paper describes how artificial neural networks can be used to classify multivariate data. Two types of neural networks were applied: a counter propagation neural network (CP-ANN) and a radial basis function network (RBFN). These strategies were used to classify soil samples from different geographical regions in Brazil by means of their near-infrared (diffuse reflectance) spectra. The results were better with CP-ANN (classification error 8.6%) than with RBFN (classification error 11.0%).     

多变量统计分类技术——磺胺甲基异口恶唑质量评价的研究

本文研究了近红外漫反射光谱法进行磺胺甲基异恶唑粉末药品质量评价的可能性，用多变量统计分类技术（系统聚类分析，逐步聚类分析，主成分分析和逐步判别分析），从磺胺甲基异恶唑末药品的一阶导数光谱，成功地鉴别了真药，劣药和假药，结果令人满意。     

Determination of organic matter in soils using radial basis function networks and near infrared spectroscopy

A relationship was established between the organic matter content in soils determined by conventional chemical measurements and by diffuse reflectance spectra in the near infrared region (1000-2500 nm). Radial basis function networks (RBFN) with regularized forward selection to control the model complexity were used for non-parametric regression, resulting in a RMSEP of 0.25%. The observed results using RBFN were better than those obtained by partial least squares regression (PLS) and multi-layer perceptron (MLP) feed-forward networks with a back-propagation learning algorithm. RBFN is a suitable tool to model this complex system, with additional advantages over MLP, since the training procedure is less dependent on the initial conditions.