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

Jie Yu,Yong-jiang Wu,Yi-yu Cheng
DOI: https://doi.org/10.3321/j.issn:1001-5302.2002.02.005
2002-01-01
Abstract: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.
What problem does this paper attempt to address?