一种基于分析数据可视化技术的中药材质量分析与评价新方法

目的 :研究中药材质量分析与评价的新方法。方法 :运用科学计算可视化原理 ,建立了仪器分析数据可视化技术。通过主成分分析和空间投影变换方法对药材的红外光谱分析数据进行预处理 ,实现特征提取和高维数据空间的降维 ,进而采用二维灰度图对变换后的数据进行可视化处理 ,获取中药材质量的特征指纹图谱。结果 :对野生、栽培桑白皮及其伪品共 42个样品进行了测定。结果表明 ,样品鉴别的准确率达到 90.5%。结论 :该方法是一种具有重要发展意义的中药材质量鉴定新技术。

A Novel Fingerprint Method for Quality Evaluation of Chinese Medicinal Plants Based on Analytical Data Visualization

Objective: To investigate new method for evaluating the quality of Chinese Medicinal Plants (CMP) . Method: A visualization technique for representing instrumental analytical data was developed by applying the fundamental of Data Visualization,with Principal Component Analysis (PCA) and spatial projection transformation, original IR spectral data were projected into a low dimensional subspace so that the dimensionality of original data space was decreased and tiny fingerprint features were extracted. The data set in the subspace was visualized by means of two dimensional grayscale images. Consequently, the characteristic fingerprint for appraising the quality of CMP was obtained. Result: 42 mulberry root bark samples from three different quality classes were identified with the proposed method, which showed that the fingerprint images had satisfactory resolution and classification accuracy as high as 90.5%. Conclusion: The proposed method is a useful technique for appraising the quality of CMP.