Cordella Christophe B Y, Militão Julio S L T, Clément Marie-Claude, Cabrol-Bass Daniel
Agence Française de Sécurité Sanitaire des Aliments, Laboratoire d'Etudes et de Recherches sur les Petits Ruminants et les Abeilles, Unité Abeille du LPPRA-Sophia Antipolis, 105 route des Chappes, F-06902 Sophia-Antipolis Cedex, France.
J Agric Food Chem. 2003 May 21;51(11):3234-42. doi: 10.1021/jf021100m.
An improved COFRAC (COmité FRançais d'ACréditation) method for the analysis and evaluation of the quality of honey by high-performance anion-exchange chromatography of sugar profiles is proposed. With this method, both minor and major sugars are simultaneously analyzed and the technique is integrated in a new chemometric approach, which uses the entire chromatographic sugars profile of each analyzed sample to characterize honey floral species. Sixty-eight authentic honey samples (6 varieties) were analyzed by high-performance anion-exchange chromatography-pulsed amperometric detection. A new algorithm was developed to create automatically the corresponding normalized data matrix, ready-to-use in various chemometric procedures. This algorithm transforms the analytical profiles to produce the corresponding calibrated table of the surfaces or intensities according to retention times of peaks. The possibility of taking into account unknown peaks (those for which no standards are available) allows the maximum chemical information provided by the chromatograms to be retained. The parallel application of principal component analysis (PCA)/linear discriminant analysis (LDA) and artificial neural networks (ANN) shows a high capability in the classification of the analyzed samples (LDA, 93%; ANN, 100%) and a very good discrimination of honey groups. This work is the starting point of the elaboration of a new system designed for the automatic pattern recognition of food samples (first application on honey samples) from chromatographic analyses for food characterization and adulteration detection.
提出了一种改进的法国认可委员会（COFRAC）方法，用于通过糖谱的高效阴离子交换色谱法分析和评估蜂蜜质量。采用该方法可同时分析主要和次要糖类，并且该技术被整合到一种新的化学计量学方法中，该方法利用每个分析样品的完整色谱糖谱来表征蜂蜜的花种。通过高效阴离子交换色谱 - 脉冲安培检测法对68个真实蜂蜜样品（6个品种）进行了分析。开发了一种新算法，用于自动创建相应的归一化数据矩阵，可直接用于各种化学计量学程序。该算法根据峰的保留时间对分析图谱进行转换，以生成相应的表面或强度校准表。考虑未知峰（那些没有标准品的峰）的可能性使得色谱图提供的最大化学信息得以保留。主成分分析（PCA）/线性判别分析（LDA）和人工神经网络（ANN）的并行应用在分析样品分类方面显示出很高的能力（LDA为93%；ANN为100%），并且对蜂蜜组有很好的区分度。这项工作是为从色谱分析中自动识别食品样品（首次应用于蜂蜜样品）以进行食品表征和掺假检测而设计的新系统开发的起点。
