A Novel Multi-Preprocessing Integration Method for the Qualitative and Quantitative Assessment of Wild Medicinal Plants: Gentiana rigescens as an Example.

Zhimin Liu,Yanli Zhao,Yuanzhong Wang,Tao Shen,Zhitian Zuo,Jinyu Zhang,Zhimin Li,Ji Zhang

doi:10.3389/fpls.2021.759248

Abstract

Until now, the over-exploitation of wild resources has increased growing concern over the quality of wild medicinal plants. This led to the necessity of developing a rapid method for the evaluation of wild medicinal plants. In this study, the content of total secoiridoids (gentiopicroside, swertiamarin, and sweroside) of Gentiana rigescens from 37 different regions in southwest China were analyzed by high performance liquid chromatography (HPLC). Furthermore, Fourier transform infrared (FT-IR) was adopted to trace the geographical origin (331 individuals) and predict the content of total secoiridoids (273 individuals). In the traditional FT-IR analysis, only one scatter correction technique could be selected from a series of preprocessing candidates to decrease the impact of the light correcting effect. Nevertheless, different scatter correction techniques may carry complementary information so that using the single scatter correction technique is sub-optimal. Hence, the emerging ensemble approach to preprocessing fusion, sequential preprocessing through orthogonalization (SPORT), was carried out to fuse the complementary information linked to different preprocessing methods. The results suggested that, compared with the best results obtained on the scatter correction modeling, SPORT increased the accuracy of the test set by 12.8% in qualitative analysis and decreased the RMSEP by 66.7% in quantitative analysis.

Highlights

The medicine plant, commonplace throughout human history, is an indispensable part of modern human medicine and traditional medicine (Balunas and Kinghorn, 2005; Ramawat et al, 2009; Bai et al, 2019)
The results indicated that sequential preprocessing through orthogonalization (SPORT), compared with the second derivative (SD) approach, significantly improved the predictive performance, FIGURE 5 | The results of five PLSR models on the calibration set and test set and their residual plot. (A) multiplicative scatter correction (MSC); (B) standard normal variate (SNV); (C) variables sorting for normalization (VSN); (D) SD
As a matter of fact, 9 latent variables (LVs), 10 LVs, and 10 LVs were extracted in data processed by MSC, SNV, and SD, respectively, in terms of qualitative analysis, and 10 LVs and 10 LVs were, respectively, extracted from data processed by MSC and SD in terms of quantitative analysis

Summary

Introduction

The medicine plant, commonplace throughout human history, is an indispensable part of modern human medicine and traditional medicine (Balunas and Kinghorn, 2005; Ramawat et al, 2009; Bai et al, 2019). With the increasing demand for high-quality wild G. rigescens, the contemporary phenomenon of origin fraud and using low-quality G. rigescens as a substitute has been frequently observed, which greatly influences the supplement of world markets and the trust of consumers. These factors have led to the necessity of applying a reliable approach for the qualitative and quantitative assessment of wild G. rigescens

Objectives

Methods

Results

Discussion

Conclusion