基于信息融合的中药多元色谱指纹图谱相似性计算方法

A Computational Method Based on Information Fusion for Evaluating the Similarity of Multiple Chromatographic Fingerprints of TCM

用信息融合算法合并多张色谱指纹图谱,解决中药多元指纹图谱相似性评价难题,提出一种多元色谱指纹图谱相似度计算方法.用该方法先对各单元指纹图谱进行串行或并行象素级信息融合,再对融合了各指纹峰信息的多元色谱指纹图谱进行相似度评价.计算机仿真和复方丹参滴丸多元HPLC指纹图谱应用结果表明,该法能够评价中药多元色谱指纹图谱相似度,定量表征中成药产品批次间质量波动情况.

An computational method based on information fusion is developed to evaluate the similarity of multiple chromatographic fingerprints of TCM, where it solves a current key problem on evaluating the similarity of multiple fingerprints. In this method, the information of individual fingerprint was combined by using serial or parallel information fusion strategy at raw data level, and then the integral similarity of multiple fingerprints was calculated by comparing their fusion results. Subsequently, this method was applied to simulated datasets and a set of Compound Danshen Dripped PiIIs（CDDP） samples. The results indicate that, by this method, the similarity among multiple chromatographic fingerprints of TCM can be calculated, and the lot-to-lot consistency of samples can be evaluated quantitatively.