中药二维信息指纹图谱模式识别中药二维信息指纹图谱模式识别

目的用二维信息数据表征中药指纹图谱样本,并对其进行模式识别。方法对一维信息数据和二维信息数据表征的中药指纹图谱进行聚类分析,并对分析结果进行比较。结果与一维信息数据相比,采用二维信息数据表征指纹图谱,各种聚类方法所得总体类别差异小,样本归属比较稳定。对未知样本的类别判定聚类分析和神经网络预测结果一致。结论二维信息数据较一维信息数据更能全面、特异地表征中药指纹图谱。

Pattern recognition applied to the fingerprint of traditional Chinese medicine characterized by two-dimension information data

Aim To establish the two-dimension information fingerprints chromatograph of tradit ional Chinese medicine and investigate via pattern recognition. Methods Hierarchical cluster was applied to fingerprints described by one-dimension inf ormation data and by two-dimension information data respectively and correspond ing results was compared. Results The gross classes, in which the samples denoted by two-dimension information da ta, classified by all kinds of hierarchical cluster methods had less differences and more robustness than the methods containing the samples expressed by one-d imension information data. The classes to which the unknown samples would be bel onged were determined by hierarchical cluster analysis and artificial neural net work (ANN), and the same results were obtained. Conclusion The fingerprints characterized by two-dimension information data could provide more overall and special information as compared with the methods indicated by o ne-dimension information data.