# Quality Assessment of Gentiana rigescens from Different Geographical Origins Using FT-IR Spectroscopy Combined with HPLC

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Result and Discussion

#### 2.1. HPLC Analysis

#### 2.2. FT-IR Spectral Features

^{−1}FT-IR spectra of different parts of G. rigescens from different geographical origins are shown in Figure 2. On the whole, there is no distinct difference among the average FT-IR spectra, which overlap. However, the absorption intensities of the average FT-IR spectra vary a lot. Compared to other geographical origins, the absorption intensity is obviously lower in the root sample of Yuxi (Figure 2A). A broad absorption band is found at around 3399 cm

^{−1}, which is due to the O–H stretch. The bands at 2933 and 2856 cm

^{−1}are CH

_{3}asymmetric stretching and stretching vibration of esters, respectively. The peak around 1937 cm

^{−1}is assigned to the C=O stretching vibration of acid amides [27]. In addition, the intense absorption peaks in 1070 and 1619 cm

^{−1}are the main absorption bands of iridoids or glycosides, which correspond to C–O or C–O–C stretching and C–C asymmetric stretching vibrations [28,29]. According to studies of G. rigescens by Mi et al. [29] and Yang et al. [30], the active compounds gentiopicroside, swertiamarin, and chiratin and other iridoids in G. rigescens all contain C–O or C–O–C and C–C bonds.

#### 2.3. Multivariate Analysis

#### 2.3.1. PLS-DA Models

^{2}), root-mean-square error of estimation (RMSEE) and root-mean-square error of cross validation (RMSECV) are listed in Table 1. The first six principal components (96.0%) are employed for model 1. The R

^{2}is greater than 0.94 and the RMSEE and RMSECV are low, which are less than 0.25. Thereinto, model 1 of samples from Yuxi have the best performance with the highest R

^{2}and lowest RMSEE and RMSECV. As seen in Table 2, according to the Galtier criterion, all the samples are identified correctly except the four samples numbered 4, 6, 13 and 57. Sample 13 from Lijiang was misidentified as coming from Diqing, and the three other samples (4, 6 and 57) can’t be judged accurately. More interestingly, the uncertain samples are all outside of the 95% confidence ellipses in the scatter plot (Figure 3A). The prediction accuracy of model 1 is 80%.

^{2}, RMSEE and RMSECV in model 2. The first four principal components are employed for model 2, and the cumulative contribution reached 88.5%. The model of samples from Yuxi have the best precision, with high R

^{2}(0.9472) and low RMSEE (0.0877) and RMCECV (0.1689).

^{2}, RMSEE and RMSECV of model 3 are shown in Table 1. The performances of the different geographical origin discrimination are good, with high R

^{2}(>0.88) and low RMSEE (<0.17) and RMSECV (<0.20). Thereinto, the best performance is for the samples from Yuxi. In Table 4, the samples 128 and 137 can’t be confirmed. Moreover, a sample from Lijiang (146) is judged as a Dali sample by mistake. More interestingly, the uncertain samples are in the margin of the 95% confidence ellipses (Figure 3C) like the result of model 1. The prediction accuracy of model 3 is 85%.

#### 2.3.2. SVM Regression Model

^{20}and 2

^{−20}to 1, respectively. As can be seen in Figure 4, the results of c, g and cross-validation mean square error (CVmse) which are calculated by the GS algorithm are 0.5, 0.0039 and 0.0149, respectively. In addition, the terminate algebra was set as 200 and population quantity was set as 40 in the GA algorithm. It is shown that the optimum parameters c, g and CVmse are 0.4572, 0.01 and 0.0163, respectively (Figure 5). Finally, the PSO algorithm was also applied to select the parameters and the detail parameter (terminate algebra and population quantity) of PSO was the same as the GA algorithm (Figure 6). The results of the PSO algorithm are as follows: c = 0.4453, g = 0.01 and CVmse = 0.01624. The aforementioned algorithms were all applied for building the SVM regression models.

_{t}

^{2}(96.39%) and RMSEE (3.1056) for training set and the highest R

_{v}

^{2}(83.57%) and the lowest RMSEP (11.1421) for validation set is obtained by the GS algorithm. Therefore, the GS method gives the best performance for the prediction of gentiopicroside content in G. rigescens.

_{t}

^{2}, R

_{t}

^{2}, RMSEE and RMSEP are achieved and good agreement with the SVM regression model built for predicting gentiopicroside content in G. rigescens is observed.

## 3. Materials and Methods

#### 3.1. Plant Materials and Reagents

#### 3.2. Sample Preparation

#### 3.3. HPLC Conditions

#### 3.4. FT-IR Spectra Acquisition

^{−1}with a resolution of 4 cm

^{−1}and a total of 16 co-added scans. Pure KBr spectra were recorded as background spectra for deducting the CO

_{2}and H

_{2}O peaks in real-time. Each spectrum was scanned in triplicate under constant temperature (25 °C) and humidity conditions, and the averaged spectra were employed for further analysis.

#### 3.5. Multivariate Data Analysis

#### 3.6. Evaluation of Model Performance

^{2}), root-mean-square error of estimation (RMSEE), root-mean-square error of cross validation (RMSECV) and root-mean-square error of prediction (RMSEP) were considered to evaluate the performance of qualitative and quantitative model.

^{2}(Equation (1)) is the correlation between the measured values and predicted values. Generally, a higher R

^{2}(<1) value means a better performance of both kinds of models [51]:

_{i}is the measured value while ŷ

_{i}is the predicted value. $\overline{\mathrm{y}}$ is the mean value, and N is the number of samples.

_{t}is the number of the training set and N

_{v}is the number of validation set. In addition, in the qualitative model, the classification accuracy of the validation set depends on the predicted value (Y

_{pre}), and deviation values (Y

_{dev}) which are based on the following standards: (1) when Y

_{pre}> 0.5 and Y

_{dev}< 0.5, the sample of validation set belongs to the class; (2) when Y

_{pre}< 0.5 and Y

_{dev}< 0.5, the sample of validation set does not belong to the class; (3) when Y

_{dev}> 0.5, it means that the sample can’t judge accurately whether it belongs to the class or not [56,57].

#### 3.7. Software

## 4. Conclusions

## Acknowledgments

## Author Contributions

## Conflicts of Interest

## References

- Licata, A.; Macaluso, F.S.; Craxì, A. Herbal hepatotoxicity: A hidden epidemic. Intern. Emerg. Med.
**2013**, 8, 13–22. [Google Scholar] [CrossRef] [PubMed] - Choi, Y.H.; Chin, Y.W.; Kim, Y.G. Herb-drug interactions: Focus on metabolic enzymes and transporters. Arch. Pharm. Res.
**2011**, 34, 1843–1863. [Google Scholar] [CrossRef] [PubMed] - Gad, H.A.; El-Ahmady, S.H.; Abou-Shoer, M.I.; Abou-Shoerb, M.I.; Al-Azizi, M.M. Application of chemometrics in authentication of herbal medicines: A review. Phytochem. Anal.
**2013**, 24, 1–24. [Google Scholar] [CrossRef] [PubMed] - Anderson, M.L.; Burney, D.P. Validation of sample preparation procedures for botanical analysis. J. AOAC Int.
**1998**, 81, 1005–1010. [Google Scholar] - Huie, C.W. A review of modern sample-preparation techniques for the extraction and analysis of medicinal plants. Anal. Bioanal. Chem.
**2002**, 373, 23–30. [Google Scholar] [CrossRef] [PubMed] - State Pharmacopoeia Commission. Chinese Pharmacopoeia; China Medical Science and Technology Press: Beijing, China, 2015; Volume I, pp. 260–261. [Google Scholar]
- Pan, Y.; Zhang, J.; Shen, T.; Zuo, Z.T.; Jin, H.; Wang, Y.Z.; Li, W.Y. Optimization of ultrasonic extraction by response surface methodology combined with ultrafast liquid chromatography–ultraviolet method for determination of four iridoids in Gentiana rigescens. J. Food Drug Anal.
**2014**, 23, 529–537. [Google Scholar] [CrossRef] - Xu, M.; Wang, D.; Zhang, Y.J.; Yang, C.R. Dammarane triterpenoids from the roots of Gentiana rigescens. J. Nat. Prod.
**2007**, 70, 880–883. [Google Scholar] [CrossRef] [PubMed] - Zhang, X.D.; Allan, A.C.; Li, C.X.; Wang, Y.Z.; Yao, Q.Y. De Novo assembly and characterization of the transcriptome of the Chinese medicinal herb, Gentiana rigescens. Int. J. Mol. Sci.
**2015**, 16, 11550–11573. [Google Scholar] [CrossRef] [PubMed] - Melito, S.; Petretto, G.L.; Podani, J.; Foddai, M.; Maldini, M.; Chessa, M.; Pintore, G. Altitude and climate influence Helichrysum italicum subsp. microphyllum essential oils composition. Ind. Crops Prod.
**2016**, 80, 242–250. [Google Scholar] [CrossRef] - Wu, Z.; Zhang, J.; Xu, F.R.; Wang, Y.Z.; Zhang, J.Y. Rapid and simple determination of polyphyllin I, II, VI, and VII in different harvest times of cultivated Paris polyphylla Smith var. yunnanensis (Franch.) Hand.-Mazz by UPLC-MS/MS and FT-IR. J. Nat. Med.
**2017**, 71, 139–147. [Google Scholar] [CrossRef] [PubMed] - Shu, X.; Jiang, X.W.; Cheng, B.C.Y.; Ma, S.C.; Chen, G.Y.; Yu, Z.L. Ultra-performance liquid chromatography-quadrupole/time-of-flight mass spectrometry analysis of the impact of processing on toxic components of Kansui Radix. BMC Complement. Altern. Med.
**2016**, 16, 73. [Google Scholar] [CrossRef] [PubMed] - Yu, W.J.; Ma, M.Y.; Chen, X.M.; Li, L.R.; Zheng, Y.F. Traditional Chinese Medicine and Constitutional Medicine in China, Japan and Korea: A Comparative Study. Am. J. Chin. Med.
**2017**, 45, 1–12. [Google Scholar] [CrossRef] [PubMed] - Jing, P.; Zhao, S.J.; Lu, M.M.; Cai, Z.; Pang, J.; Song, L.H. Multiple-fingerprint analysis for investigating quality control of Flammulina velutipes fruiting body polysaccharides. J. Agric. Food Chem.
**2014**, 62, 12128–12133. [Google Scholar] [CrossRef] [PubMed] - Choong, Y.K.; Xu, C.H.; Lan, J.; Chen, X.D.; Jamal, J.A. Identification of geographical origin of Lignosus samples using Fourier transform infrared and two-dimensional infrared correlation spectroscopy. J. Mol. Struct.
**2014**, 1069, 188–195. [Google Scholar] [CrossRef] - Xia, Y.Y.; Liu, F.L.; Feng, F.; Liu, W.Y. Characterization, quantitation and similarity evaluation of Codonopsis lanceolata from different regions in China by HPLC-Q-TQF-MS and chemometrics. J. Food Compos. Anal.
**2017**, 62, 134–142. [Google Scholar] [CrossRef] - Kaniu, M.I.; Angeyo, K.H.; Mwala, A.K.; Mangala, M.J. Direct rapid analysis of trace bioavailable soil macronutrients by chemometrics-assisted energy dispersive X-ray fluorescence and scattering spectrometry. Anal. Chim. Acta
**2012**, 729, 21–25. [Google Scholar] [CrossRef] [PubMed] - Shao, Q.S.; Zhang, A.; Ye, W.W.; Guo, H.P.; Hu, R.H. Fast determination of two atractylenolides in Rhizoma Atractylodis Macrocephalae by Fourier transform near-infrared spectroscopy with partial least squares. Spectrochim. Acta A Mol. Biomol. Spectrosc.
**2014**, 120, 499–504. [Google Scholar] [CrossRef] [PubMed] - Chavez, P.F.; Sacre, P.Y.; De Bleye, C.; Netchacovitch, L.; Mantanus, J.; Motteb, H.; Schubert, M.; Hubert, P.; Ziemons, E. Active content determination of pharmaceutical tablets using near infrared spectroscopy as Process Analytical Technology tool. Talanta
**2015**, 144, 1352–1359. [Google Scholar] [CrossRef] [PubMed] - Feng, Y.C.; Lei, D.Q.; Hu, C.Q. Rapid identification of illegal synthetic adulterants in herbal anti-diabetic medicines using near infrared spectroscopy. Spectrochim Acta A Mol. Biomol. Spectrosc.
**2014**, 125, 363–374. [Google Scholar] [CrossRef] [PubMed] - Zhao, Y.L.; Zhang, J.; Jin, H.; Zhang, J.Y.; Shen, T.; Wang, Y.Z. Discrimination of Gentiana rigescens from different origins by Fourier transform infrared spectroscopy combined with chemometric methods. J. AOAC Int.
**2015**, 98, 22–26. [Google Scholar] [CrossRef] [PubMed] - Qi, L.M.; Zhang, J.; Zuo, Z.T.; Zhao, Y.L.; Wang, Y.Z.; Jin, H. Determination of Iridoids in Gentiana rigescens by Infrared Spectroscopy and Multivariate Analysis. Anal. Lett.
**2017**, 50, 389–401. [Google Scholar] [CrossRef] - Qi, L.M.; Zhang, J.; Zhao, Y.L.; Zou, Z.T.; Jin, H.; Wang, Y.Z. Quantitative and Qualitative Characterization of Gentiana rigescens Franch (Gentianaceae) on Different Parts and Cultivations Years by HPLC and FTIR Spectroscopy. J. Anal. Methods Chem.
**2017**, 2017, 3194146. [Google Scholar] [CrossRef] [PubMed] - Shen, T.; Zhang, J.; Zhao, Y.L.; Zuo, Z.T.; Wang, Y.Z. Chemometric Analysis of the Stem and Leaf of Gentiana rigescens in Agroforestry Systems. Plant Sci. J.
**2015**, 33, 472–481. [Google Scholar] - Cheng, J.G.; Wang, X.F.; Fan, L.Z.; Yang, X.P.; Yang, P.W. Variations of Yunnan climatic zones in recent 50 years. Prog. Geogr.
**2009**, 28, 18–24. [Google Scholar] - Zhao, Y.L.; Zhang, J.; Yuan, T.J.; Shen, T.; Li, W.; Yang, S.H.; Hou, Y.; Wang, Y.Z.; Jin, H. Discrimination of wild Paris based on near infrared spectroscopy and high performance liquid chromatography combined with multivariate analysis. PLoS ONE
**2014**, 9, e89100. [Google Scholar] [CrossRef] [PubMed] - Cheng, C.G.; Liu, J.; Cao, W.Q.; Zheng, R.W.; Wang, H.; Zhang, C.J. Classification of two species of Bidens based on discrete stationary wavelet transform extraction of FTIR spectra combined with probability neural network. Vib. Spectrosc.
**2010**, 54, 50–55. [Google Scholar] [CrossRef] - Xu, M.; Wang, D.; Zhang, Y.J.; Yang, C.R. A new secoiridoidal glucoside from Gentiana rigescens (Gentianaceae). Acta Bot. Yunnanica
**2006**, 28, 669–672. [Google Scholar] - Mi, L.J.; Zhang, J.; Zhao, Y.L.; Wang, Y.Z.; Li, F.S. Application of infrared spectroscopy for choosing drying methods of Gentiana rigescens. Lishizhen Med. Mater. Med. Res.
**2015**, 26, 2656–2659. [Google Scholar] [CrossRef] - Yang, H.X.; Ma, F.; Du, Y.Z.; Sun, S.Q.; Wei, L.X. Study on the Tiabetan medicine Swertia mussotii Franch and its extracts by Fourier transform infrared spectroscopy. Spectrosc. Spectr. Anal.
**2014**, 34, 2973–2977. [Google Scholar] - Kennard, R.W.; Stone, L.A. Computer aided design of experiments. Technometrics
**1969**, 11, 137–148. [Google Scholar] [CrossRef] - Rinnan, Å.; van den Berg, F.; Engelsen, S.B. Review of the most common pre-processing techniques for near-infrared spectra. TRAC Trends Anal. Chem.
**2009**, 28, 1201–1222. [Google Scholar] [CrossRef] - Maleki, M.R.; Mouazen, A.M.; Ramon, H.; De Baerdemaeker, J. Multiplicative scatter correction during on-line measurement with near infrared spectroscopy. Biosyst. Eng.
**2007**, 96, 427–433. [Google Scholar] [CrossRef] - Guo, Q.; Wu, W.; Massart, D.L. The robust normal variate transform for pattern recognition with near-infrared data. Anal. Chim. Acta
**1999**, 382, 87–103. [Google Scholar] [CrossRef] - Hsu, C.W.; Chang, C.C.; Lin, C.J. A Practical Guide to Support Vector Classification. 2003. Available online: http://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf (accessed on 10 May 2017).
- Kalteh, A.M. Wavelet genetic algorithm-support vector regression (Wavelet GA-SVR) for monthly flow forecasting. Water Resour. Manag.
**2015**, 29, 1283–1293. [Google Scholar] [CrossRef] - Cheng, Z.Y.; Kong, H.H.; Zhang, J.; Bai, W.L.; Gan, F. Application of Particle Swarm Optimization-Least Square Support Vector Machine Regression to Modeling of Near Infrared Spectra. J. Instrum. Anal.
**2010**, 12, 1215–1219. [Google Scholar] [CrossRef] - Grasel, F.S.; Ferrão, M.F. A rapid and non-invasive method for the classification of natural tannin extracts by near-infrared spectroscopy and PLS-DA. Anal. Methods
**2016**, 8, 644–649. [Google Scholar] [CrossRef] - Górski, Ł.; Sordoń, W.; Ciepiela, F.; Kubiak, W.W.; Jakubowska, M. Voltammetric classification of ciders with PLS-DA. Talanta
**2016**, 146, 231–236. [Google Scholar] [CrossRef] [PubMed] - Dominguez-Vidal, A.; Pantoja-de la Rosa, J.; Cuadros-Rodríguez, L.; Ayora-Cañada, M.J. Authentication of canned fish packing oils by means of Fourier transform infrared spectroscopy. Food Chem.
**2016**, 190, 122–127. [Google Scholar] [CrossRef] [PubMed] - Cortes, C.; Vapnik, V. Support-vector networks. Mach. Learn.
**1995**, 20, 273–297. [Google Scholar] [CrossRef] - Yang, Y.; Liu, X.S.; Li, W.L.; Jin, Y.; Wu, Y.J.; Zheng, J.Y.; Zhang, W.T.; Chen, Y. Rapid measurement of epimedin A, epimedin B, epimedin C, icariin, and moisture in Herba Epimedii using near infrared spectroscopy. Spectrochim. Acta A Mol. Biomol. Spectrosc.
**2017**, 171, 351–360. [Google Scholar] [CrossRef] [PubMed] - Chen, Q.S.; Zhao, J.W.; Fang, C.H.; Wang, D.M. Feasibility study on identification of green, black and Oolong teas using near-infrared reflectance spectroscopy based on support vector machine (SVM). Spectrochim. Acta A Mol. Biomol. Spectrosc.
**2007**, 66, 568–574. [Google Scholar] [CrossRef] [PubMed] - Gunn, S.R. Support Vector Machines for Classification and Regression. ISIS Tech. Rep.
**1998**, 14, 85–86. [Google Scholar] - Zhang, Y.; Cong, Q.; Xie, Y.F.; Yang, J.X.; Zhao, B. Quantitative analysis of routine chemical constituents in tobacco by near-infrared spectroscopy and support vector machine. Spectrochim. Acta A Mol. Biomol. Spectrosc.
**2008**, 71, 1408–1413. [Google Scholar] [CrossRef] [PubMed] - Chang, C.C.; Lin, C.J. LIBSVM: A library for support vector machines. ACM Trans. Intell. Syst. Technol.
**2011**, 2, 27. [Google Scholar] [CrossRef] - Subasi, A. Classification of EMG signals using PSO optimized SVM for diagnosis of neuromuscular disorders. Comput. Biol. Med.
**2013**, 43, 576–586. [Google Scholar] [CrossRef] [PubMed] - Benkedjouh, T.; Medjaher, K.; Zerhouni, N.; Rechak, S. Health assessment and life prediction of cutting tools based on support vector regression. J. Intell. Manuf.
**2015**, 26, 213–223. [Google Scholar] [CrossRef] [Green Version] - Ulenberg, S.; Belka, M.; Król, M.; Herold, F.; Hewelt-Belka, W.; Kot-Wasik, A.; Bączek, T. Prediction of overall in vitro microsomal stability of drug candidates based on molecular modeling and support vector machines. Case study of novel arylpiperazines derivatives. PLoS ONE
**2015**, 10, e0122772. [Google Scholar] [CrossRef] [PubMed] - Liu, M.; Wang, M.; Wang, J.; Dou, L. Comparison of random forest, support vector machine and back propagation neural network for electronic tongue data classification: Application to the recognition of orange beverage and Chinese vinegar. Sens. Actuators B Chem.
**2013**, 177, 970–980. [Google Scholar] [CrossRef] - Zhong, J.F.; Qin, X.L. Rapid quantitative analysis of corn starch adulteration in Konjac Glucomannan by chemometrics-assisted FT-NIR spectroscopy. Food Anal. Methods
**2016**, 9, 61–67. [Google Scholar] [CrossRef] - Golic, M.; Walsh, K.B. Robustness of calibration models based on near infrared spectroscopy for the in-line grading of stonefruit for total soluble solids content. Anal. Chim. Acta
**2006**, 555, 286–291. [Google Scholar] [CrossRef] - Pande, R.; Mishra, H.N. Fourier transform near-infrared spectroscopy for rapid and simple determination of phytic acid content in green gram seeds (Vigna radiata). Food Chem.
**2015**, 172, 880–884. [Google Scholar] [CrossRef] [PubMed] - Bekiaris, G.; Triolo, J.M.; Peltre, C.; Pedersen, L.; Jensen, L.S.; Bruun, S. Rapid estimation of the biochemical methane potential of plant biomasses using Fourier transform mid-infrared photoacoustic spectroscopy. Bioresour. Technol.
**2015**, 197, 475–481. [Google Scholar] [CrossRef] [PubMed] - Chia, K.S.; Rahim, H.A.; Rahim, R.A. Neural network and principal component regression in non-destructive soluble solids content assessment: A comparison. J. Zhejiang Univ. Sci. B
**2012**, 13, 145–151. [Google Scholar] [CrossRef] [PubMed] - De Girolamo, A.; Lippolis, V.; Nordkvist, E.; Visconti, A. Rapid and non-invasive analysis of deoxynivalenol in durum and common wheat by Fourier-Transform Near Infrared (FT-NIR) spectroscopy. Food Addit. Contam.
**2009**, 26, 907–917. [Google Scholar] [CrossRef] [PubMed] - Shi, F.C.; Li, D.L.; Feng, G.L.; Song, G.F.; Zhou, Z.G. Discrimination of producing areas of flue-cured tobacco leaves with near infrared spectroscopy based PLS-DA algorithm. Tob. Chem.
**2013**, 4, 56–59. [Google Scholar] - Li, Y. LIBSVM-Faruto Ultimate: A Toolbox with Implements for Support Vector Machines Based on Libsvm. 2011. Available online: http://www.matlabsky.com (accessed on 10 June 2016).

Sample Availability: Not available. |

**Figure 1.**Contents of gentiopicroside in G. rigescens (mg/g) with different parts of plants from different geographical origins by HPLC.

**Figure 2.**The average FT-IR spectra of root (

**A**), stem (

**B**) and leaf (

**C**) in G. rigescens from different geographical origins (Dali, Lijiang, Diqing and Yuxi) in the 4000–400 cm

^{−1}range.

**Figure 3.**The scatter plot of PLS-DA FT-IR spectra display the information of samples of root (

**A**), stem (

**B**), leaf (

**C**) and three parts (root, stem and leaf) (

**D**) in G. rigescens from different geographical origins (Dali, Lijiang, Diqing and Yuxi). The abscissa represents the variation of the first component and the ordinate represents the variation of the second component.

**Figure 4.**The 3D view of the optimization results for parameters c and g by grid search method with seven-fold cross validation.

**Figure 7.**Correlation diagram between FT-IR predicted values and the reference values in the training and validation sets for gentiopicroside.

Types Parameter | Model 1 | Model 2 | Model 3 | Model 4 | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

Dali, Root | Diqing, Root | Lijiang, Root | Yuxi, Root | Dali, Stem | Diqing, Stem | Lijiang, Stem | Yuxi, Stem | Dali, Leaf | Diqing, Leaf | Lijiang, Leaf | Yuxi, Leaf | Root | Stem | Leaf | |

R^{2} | 0.9543 | 0.9654 | 0.9426 | 0.9762 | 0.8346 | 0.8464 | 0.9098 | 0.9472 | 0.9031 | 0.8964 | 0.8854 | 0.9231 | 0.8334 | 0.8425 | 0.9331 |

RMSEE | 0.0852 | 0.0985 | 0.1247 | 0.0615 | 0.1552 | 0.1999 | 0.1531 | 0.0877 | 0.1364 | 0.1695 | 0.1464 | 0.1140 | 0.1835 | 0.1927 | 0.1295 |

RMSECV | 0.1785 | 0.2313 | 0.1992 | 0.0709 | 0.2740 | 0.2688 | 0.1807 | 0.1689 | 0.1614 | 0.1924 | 0.1680 | 0.1713 | 0.2272 | 0.2304 | 0.1445 |

^{2}: determination coefficient; RMSEE: root-mean-square error of estimation; RMSECV: root-mean-square error of cross-validation.

Samples | Actual Class | Calculated Class | Y_{Pre} | Y_{dev} |
---|---|---|---|---|

2 | Dali, Root | Dali, Root | 1.1112 | 0.4306 |

4 | Dali, Root | Uncertain | 1.5243 | 0.6372 |

6 | Dali, Root | Uncertain | 1.5774 | 0.6637 |

7 | Dali, Root | Dali, Root | 1.2030 | 0.4765 |

13 | Lijiang, Root | Diqing, Root | 0.4190 | 0.1300 |

14 | Lijiang, Root | Lijiang, Root | 0.6108 | 0.2930 |

18 | Lijiang, Root | Lijiang, Root | 0.8904 | 0.3202 |

20 | Lijiang, Root | Lijiang, Root | 0.8029 | 0.3449 |

21 | Lijiang, Root | Lijiang, Root | 1.1986 | 0.4743 |

24 | Lijiang, Root | Lijiang, Root | 1.0592 | 0.4046 |

25 | Lijiang, Root | Lijiang, Root | 0.8340 | 0.2920 |

31 | Diqing, Root | Diqing, Root | 0.7014 | 0.2522 |

40 | Diqing, Root | Diqing, Root | 0.8934 | 0.3217 |

42 | Diqing, Root | Diqing, Root | 0.7153 | 0.2364 |

44 | Diqing, Root | Diqing, Root | 0.7702 | 0.2601 |

45 | Diqing, Root | Diqing, Root | 1.0444 | 0.3972 |

47 | Diqing, Root | Diqing, Root | 0.8626 | 0.3079 |

54 | Yuxi, Root | Yuxi-Root | 0.9195 | 0.3351 |

57 | Yuxi, Root | Uncertain | 1.3332 | 0.5416 |

59 | Yuxi, Root | Yuxi-Root | 1.1181 | 0.4341 |

_{pre}: predicted value; Y

_{dev}and deviation values.

Samples | Actual Class | Calculated Class | Y_{Pre} | Y_{dev} |
---|---|---|---|---|

61 | Dali, Stem | Dali, Stem | 0.9741 | 0.4110 |

62 | Dali, Stem | Dali, Stem | 1.1943 | 0.4722 |

66 | Dali, Stem | Dali, Stem | 0.8053 | 0.2776 |

70 | Dali, Stem | Dali, Stem | 0.5274 | 0.1387 |

72 | Lijiang, Stem | Lijiang, Stem | 1.0518 | 0.4318 |

74 | Lijiang, Stem | Lijiang, Stem | 0.9547 | 0.3524 |

76 | Lijiang, Stem | Uncertain | 1.3367 | 0.6355 |

81 | Lijiang, Stem | Lijiang, Stem | 0.7931 | 0.2796 |

87 | Lijiang, Stem | Lijiang, Stem | 1.0656 | 0.4078 |

88 | Lijiang, Stem | Lijiang, Stem | 1.0032 | 0.3766 |

91 | Diqing, Stem | Diqing, Stem | 1.2258 | 0.4879 |

94 | Diqing, Stem | Diqing, Stem | 1.1724 | 0.4612 |

95 | Diqing, Stem | Diqing, Stem | 1.0654 | 0.4077 |

104 | Diqing, Stem | Lijiang, Stem | 0.4763 | 0.2348 |

105 | Diqing, Stem | Diqing, Stem | 0.6590 | 0.2045 |

107 | Diqing, Stem | Lijiang, Stem | 0.3148 | 0.2175 |

112 | Yuxi, Stem | Yuxi, Stem | 0.9311 | 0.3704 |

117 | Yuxi, Stem | Yuxi, Stem | 0.9702 | 0.3601 |

120 | Yuxi, Stem | Yuxi, Stem | 0.9334 | 0.3417 |

_{pre}: predicted value; Y

_{dev}and deviation values.

Samples | Actual Class | Calculated Class | Y_{Pre} | Y_{dev} |
---|---|---|---|---|

123 | Dali, Leaf | Dali, Leaf | 0.7455 | 0.2478 |

128 | Dali, Leaf | Uncertain | 1.4334 | 0.5917 |

133 | Lijiang, Leaf | Lijiang, Leaf | 0.8984 | 0.3242 |

134 | Lijiang, Leaf | Lijiang, Leaf | 0.6928 | 0.2214 |

135 | Lijiang, Leaf | Lijiang, Leaf | 0.5073 | 0.1949 |

136 | Lijiang, Leaf | Lijiang, Leaf | 0.8768 | 0.3134 |

137 | Lijiang, Leaf | Uncertain | 1.1790 | 0.6497 |

139 | Lijiang, Leaf | Lijiang, Leaf | 0.9106 | 0.3494 |

140 | Lijiang, Leaf | Lijiang, Leaf | 0.5083 | 0.1241 |

141 | Lijiang, Leaf | Lijiang, Leaf | 0.9160 | 0.3330 |

144 | Lijiang, Leaf | Lijiang, Leaf | 1.0707 | 0.4103 |

145 | Lijiang, Leaf | Lijiang, Leaf | 1.3312 | 0.5406 |

146 | Lijiang, Leaf | Dali, Leaf | 0.3026 | 0.1056 |

154 | Diqing, Leaf | Diqing, Leaf | 1.1157 | 0.4329 |

160 | Diqing, Leaf | Diqing, Leaf | 0.8556 | 0.3242 |

165 | Diqing, Leaf | Diqing, Leaf | 0.5760 | 0.1630 |

166 | Diqing, Leaf | Diqing, Leaf | 1.1611 | 0.4555 |

170 | Yuxi, Leaf | Yuxi, Leaf | 0.9046 | 0.3273 |

176 | Yuxi, Leaf | Yuxi, Leaf | 0.8712 | 0.3127 |

178 | Yuxi, Leaf | Yuxi, Leaf | 0.9259 | 0.3379 |

_{pre}: predicted value; Y

_{dev}and deviation values.

Samples | Actual Class | Calculated Class | YPre | Ydev |
---|---|---|---|---|

5 | Root | Root | 0.7803 | 0.3378 |

6 | Root | Root | 0.9955 | 0.4414 |

7 | Root | Root | 0.7524 | 0.2794 |

12 | Root | Root | 0.9642 | 0.4205 |

13 | Root | Root | 0.9584 | 0.4167 |

14 | Root | Root | 0.8682 | 0.3566 |

15 | Root | Root | 0.7034 | 0.2467 |

18 | Root | Root | 0.8453 | 0.3500 |

19 | Root | Root | 0.8823 | 0.3660 |

20 | Root | Root | 0.8876 | 0.3695 |

21 | Root | Root | 0.6858 | 0.2350 |

22 | Root | Root | 0.8692 | 0.3573 |

24 | Root | Root | 0.9851 | 0.4345 |

29 | Root | Uncertain | 1.1444 | 0.5407 |

30 | Root | Root | 0.9511 | 0.4118 |

31 | Root | Root | 0.6869 | 0.2357 |

32 | Root | Root | 0.9623 | 0.4193 |

34 | Root | Root | 0.8161 | 0.3218 |

38 | Root | Root | 0.9635 | 0.4201 |

40 | Root | Root | 1.0701 | 0.4912 |

41 | Root | Root | 0.6536 | 0.2135 |

43 | Root | Root | 0.7605 | 0.2848 |

49 | Root | Root | 0.9819 | 0.4324 |

50 | Root | Root | 0.7037 | 0.3252 |

51 | Root | Uncertain | 1.2548 | 0.6143 |

52 | Root | Root | 0.9847 | 0.4343 |

53 | Root | Root | 0.8527 | 0.3462 |

57 | Root | Root | 0.8420 | 0.3391 |

58 | Stem | Stem | 0.5076 | 0.1588 |

61 | Stem | Uncertain | 1.1209 | 0.5250 |

62 | Stem | Uncertain | 1.2384 | 0.6033 |

63 | Stem | Root | 0.2768 | 0.2470 |

70 | Stem | Stem | 0.6177 | 0.1896 |

71 | Stem | Stem | 0.7688 | 0.2903 |

73 | Stem | Stem | 0.6066 | 0.1822 |

78 | Stem | Root | 0.1209 | 0.2604 |

82 | Stem | Stem | 0.7153 | 0.2596 |

83 | Stem | Root | 0.4597 | 0.0842 |

84 | Stem | Stem | 0.6782 | 0.2299 |

86 | Stem | Stem | 1.0495 | 0.4932 |

89 | Stem | Stem | 0.7826 | 0.2995 |

104 | Stem | Stem | 0.7306 | 0.2648 |

106 | Stem | Uncertain | 1.2252 | 0.5946 |

108 | Stem | Stem | 0.8728 | 0.3596 |

110 | Stem | Stem | 0.7044 | 0.2639 |

111 | Stem | Uncertain | 1.5978 | 0.8430 |

112 | Stem | Stem | 1.0158 | 0.4550 |

122 | Leaf | Leaf | 1.0092 | 0.4505 |

123 | Leaf | Leaf | 0.9702 | 0.4246 |

129 | Leaf | Leaf | 0.6900 | 0.2377 |

130 | Leaf | Leaf | 0.9760 | 0.4285 |

132 | Leaf | Leaf | 0.7933 | 0.3066 |

133 | Leaf | Leaf | 0.8667 | 0.3556 |

141 | Leaf | Leaf | 0.8623 | 0.3526 |

149 | Leaf | Leaf | 0.9041 | 0.3805 |

153 | Leaf | Leaf | 0.9990 | 0.4438 |

159 | Leaf | Leaf | 0.8485 | 0.3434 |

163 | Leaf | Leaf | 0.9792 | 0.4325 |

165 | Leaf | Leaf | 1.0081 | 0.4499 |

170 | Leaf | Leaf | 0.9311 | 0.3985 |

_{pre}: predicted value; Y

_{dev}and deviation values.

Model | c | g | CVmse | R_{t}^{2} (%) | RMSEE | R_{v}^{2} (%) | RMSEP |
---|---|---|---|---|---|---|---|

GS-SVM | 0.5000 | 0.0040 | 0.0149 | 92.7143 | 3.1056 | 83.5721 | 11.1421 |

GA-SVM | 0.4573 | 0.0100 | 0.0163 | 96.3977 | 3.1760 | 82.3279 | 11.1504 |

PSO-SVM | 0.4454 | 0.0100 | 0.0162 | 96.3120 | 3.2131 | 82.3529 | 11.1506 |

_{t}

^{2}: determination coefficient for training set; R

_{v}

^{2}: determination coefficient for validated set; RMSEE: root-mean-square error of estimation; RMSEP: root-mean-square error of prediction.

No. | Site | Description | No. | Site | Description | No. | Site | Description |
---|---|---|---|---|---|---|---|---|

1–10 | Dali, Yunnan | Root | 61–70 | Dali, Yunnan | Stem | 121–130 | Dali, Yunnan | Leaf |

11–30 | Lijiang, Yunnan | Root | 71–90 | Lijiang, Yunnan | Stem | 131–149 | Lijiang, Yunnan | Leaf |

31–50 | Diqing, Yunnan | Root | 91–110 | Diqing, Yunnan | Stem | 150–169 | Diqing, Yunnan | Leaf |

51–60 | Yuxi, Yunnan | Root | 111–120 | Yuxi, Yunnan | Stem | 170–179 | Yuxi, Yunnan | Leaf |

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Wu, Z.; Zhao, Y.; Zhang, J.; Wang, Y.
Quality Assessment of *Gentiana rigescens* from Different Geographical Origins Using FT-IR Spectroscopy Combined with HPLC. *Molecules* **2017**, *22*, 1238.
https://doi.org/10.3390/molecules22071238

**AMA Style**

Wu Z, Zhao Y, Zhang J, Wang Y.
Quality Assessment of *Gentiana rigescens* from Different Geographical Origins Using FT-IR Spectroscopy Combined with HPLC. *Molecules*. 2017; 22(7):1238.
https://doi.org/10.3390/molecules22071238

**Chicago/Turabian Style**

Wu, Zhe, Yanli Zhao, Ji Zhang, and Yuanzhong Wang.
2017. "Quality Assessment of *Gentiana rigescens* from Different Geographical Origins Using FT-IR Spectroscopy Combined with HPLC" *Molecules* 22, no. 7: 1238.
https://doi.org/10.3390/molecules22071238