|Year : 2021 | Volume
| Issue : 1 | Page : 104-110
Discrimination of five species of Panax genus and their geographical origin using electronic tongue combined with chemometrics
Li-Xia Tian1, Jin-Hua Li1, Li Zhang2, Bashir Ahmad3, Lin-Fang Huang1
1 Engineering Research Center of Chinese Medicine Resource, Ministry of Education, Institute of Medicinal Plant Development, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China
2 College of Science, Sichuan Agricultural University, Yaan, Sichuan, China
3 Centre for Biotechnology and Microbiology, University of Peshawar, Pakistan
|Date of Submission||14-Jun-2020|
|Date of Acceptance||03-Aug-2020|
|Date of Web Publication||04-Feb-2021|
Prof. Lin-Fang Huang
Institute of Medicinal Plant Development, No. 151 Malianwa North Road, Haidian District, Beijing 100193
Source of Support: None, Conflict of Interest: None
Objective: Authentication is vital to the reduction of the misuse of Panax species due to their extensive array of uses and similarities between species. However, the current authentication approach is time-consuming, laborious, and costly. The aim of this study is to discriminate the botanical origins of five species in Panax genus by a rapid and simple approach. Methods: Here, an electronic tongue (E-tongue) was applied to discriminate the botanical origins of five species of Panax, i.e., Panax quinquefolius, Panax japonicus, P. japonicus var. major, Panax zingiberensis, and Panax notoginseng (representative high-, middle-, and low-latitude plants), and the four geographical origins of P.japonicus and P. japonicus var. major plants. Data preprocessing methods, including principal component analysis (PCA), hierarchical cluster analysis (HCA), and linear discriminant analysis (LDA), were used. Results: Three models can discriminate five species of Panax genus and four plants of P. japonicus and P. japonicus var. major from different geographical origins. LDA was superior to PCA and HCA in terms of satisfactory classification. Conclusion: The findings confirmed the potential of the E-tongue for performing rapid, simple, and cost-effective discrimination via LDA.
Keywords: Discrimination, electronic tongue, hierarchical cluster analysis, linear discriminant analysis, Panax genus, principal component analysis
|How to cite this article:|
Tian LX, Li JH, Zhang L, Ahmad B, Huang LF. Discrimination of five species of Panax genus and their geographical origin using electronic tongue combined with chemometrics. World J Tradit Chin Med 2021;7:104-10
|How to cite this URL:|
Tian LX, Li JH, Zhang L, Ahmad B, Huang LF. Discrimination of five species of Panax genus and their geographical origin using electronic tongue combined with chemometrics. World J Tradit Chin Med [serial online] 2021 [cited 2021 Apr 21];7:104-10. Available from: https://www.wjtcm.net/text.asp?2021/7/1/104/310939
| Introduction|| |
Ginseng herbs, the roots and rhizomes of the Panax species (Araliaceae), have a history of thousands of years as herbal medicines and food and are well-known worldwide. The major compounds of Panax species are saponins, steroids, flavonoids, phenolic glycosides, and amino acids. Modern pharmacology research indicated that they possess anti-fatigue, antitumor, antithrombotic, anti-inflammatory, anti-oxidative, and immune-enhancing effects.,, The high-latitude and low-latitude plants, Panax quinquefolius, and Panax notoginseng are the best-known species and have been subjected to various research works., The other mid- and low-latitude Panax species, Panax japonicus, P. japonicus var. major, and Panax zingiberensis have attracted less attention. P. japonicus and P. japonicus var. major have been recorded in the Chinese Pharmacopoeia; P. japonicus was also recorded in the Japanese Pharmacopoeia., Even though the five species all belong to the Panax genus, their pharmacological activities and clinical indications are significantly different due to the diversity of their chemical constituents.,,, Inappropriate identification of Panax species may happen due to their extensive array of uses, high degree of market power, and similarities between species.
The authentication of the Panax species holds the key to ensuring the quality, safety, and efficacy of the medication. Moreover, Daodi medicine is a unique concept in traditional Chinese medicine. It is also known as a genuine medicinal herb, which means that medicinal herbs growing in a specific place under its suitable ecoclimatic environment exhibit high yield and excellent quality. Therefore, it is of considerable significance to discriminate the different geographic origins of the same species. Many studies on botanical and geographical origin discrimination have been carried out. Currently, high-performance liquid chromatography, ultra-performance liquid chromatography-mass spectrometry, chromatographic fingerprints, metabolomics, and DNA barcoding are widely used to find out (dis) similarity of herbs.,,, However, these methodologies are time-consuming, require nonportable and high-cost equipment needing strictly controlled operating conditions and highly qualified trained staff, and usually require elaborate sample preparation and toxic organic reagents.
With the development of sensor systems, electronic sensory evaluation has been beneficial to many fields, including plant disease diagnostics, agriculture food taste assessment, and the quality control of traditional Chinese medicine., Electronic tongue (E-tongue) systems are emerging as a rapid, uncomplicated, and cost-effective measurement technique and easy-to-handle tool that is extremely promising for the quality evaluation of traditional Chinese medicine. E-tongue systems are formed by a set of chemical sensors with cross-sensitivity to the compounds of interest combined with data processing tools. This device can mimic the human sense of taste, detecting flavors by electric signals generated with potentiometric variations. Furthermore, it gathers global information on the analyzed solution. This step is accomplished by the statistical software that can translate sensor data into taste patterns.,,,,,,
For these reasons, the primary objective of the present study was to assess the capability of E-tongue to discriminate the botanical origins of five species in Panax genus ( P. zingiberensis, P. quinquefolius, P. notoginseng, P. japonicus var. major, and P. japonicus) and four geographical origins of P. japonicus and P. japonicus var. major. The data points generated from the current response of the E-tongue are considered for pattern recognition. Principal component analysis (PCA), hierarchical cluster analysis (HCA), and linear discriminant analysis (LDA) were used. This research provided a simple, rapid, and cost-effective approach for classifying the authentication of five Panax genus species and identifying the geographical origins of P. japonicus var. major and P. japonicus.
| Methods|| |
Samples and reagents
Five species belonging to the Panax genus, P. japonicus, P. japonicus var. major, P. zingiberensis, P. quinquefolius, and P. notoginseng were collected from different places all over the world [Table 1]. All of the samples were authenticated by Professor Linfang Huang in the Institute of Medicinal Plant Development, Chinese Academy of Medical Sciences, and Peking Union Medical College, Beijing, China.
The fresh rhizome of plants was washed and dried at 60°C in a laboratory oven. Dried samples were crushed and sieved through a 65-mesh sieve to obtain uniform particle sizes. Filtered powder (2.0 g) was placed into a 100-mL tapered bottle with 50-mL purified water and extracted with ultrasonic at 40°C for 45 min (ultrasonic power: 180 w and frequency: 40 kHz). The solution was centrifuged, and the supernatant (30 mL) was collected for analysis.
The α-ASTREE E-tongue system (Alpha M. O. S., Toulouse, France) consisted of a hexadecimal autosampler, an Ag/AgCl reference electrode, a data acquisition system, a workstation, and seven sensors, namely, ZZ2808-2-512, CA2804-2-440, DA2808-12-330, BA2808-2-230, GA2808-2-361, BB2011-09-141, and AB2011-10-010, which were abbreviated to ZZ, CA, DA, BA, GA, BB, and AB, respectively. The seven sensors were developed to mimic the human tongue, which recognizes the overall taste information on the analyzed solution. The sensor array was immersed in the sample solution, and the response signals at the equilibrium state were collected as variables for statistical analysis. Each sample was measured for 120 s, and the signal data were recorded every 1 s. Between measurement cycles, distilled water was applied to wash the sensors for 30 s to ensure stable potentials. Based on the results of preliminary experiments and the need for a stability system, an E-tongue measurement was performed seven times for each sample. The data acquired in the latter three times were used for subsequent chemometric analysis. Each assay was carried out in triplicate. All measurements were performed at room temperature (25°C ± 1°C).
With the development of statistical software, many problems have been addressed using a wide variety of mathematical/statistical analyses. PCA, HCA, and LDA were applied to analyze electrochemical responses.
PCA is a very well-known unsupervised multivariate statistical method that reduces the dimensionality of data while retaining the important variation in the data; it is widely applied in many fields.,, The PCA expresses information contained in a dataset by a lower number of variables called principal components. These principal components are linear combinations of the original response vectors. The representative components are selected to include the maximum data variance and to be orthogonal. Hence, PCA allows the reduction of multidimensional data to a lower-dimensional approximation while simplifying the interpretation of the data by the first two or three principal components (PC1, PC2, and PC3) in two or three dimensions and preserving most of the variance in the data.
HCA is a classification tool based on the similarities among the samples, and individuals within the same group are more similar to each other than those belonging to different groups. The clustering of the samples is obtained by calculating the Euclidean distance between them followed by stepwise clustering of the most similar variables. The clusters' result is typically presented through a dendrogram. The shorter distance between two objects in the dendrograms means higher similarities between them. In this study, Ward's linkage criterion was selected. The analyses were performed using the statistical software RStudio.
LDA is a prevalent and simple supervised method for data discrimination and dimensionality reduction. It is one of the most favored classification approaches according to selecting the class sharing the highest posterior probability. LDA makes full use of the label information to learn a discriminant projection that can significantly maximize the between-class distance and minimize the within-class distance to guarantee the classification's accuracy.
| Results|| |
The sensor array analysis
Seven sensors, namely, ZZ, JE, BB, CA, GA, HA, and JE, were used to measure the samples of five species belonging to the Panax genus and two species which were further studied for their taste characters from different producing areas. The seven sensors adsorbed diverse ionized molecules in the solution selectively because of various detected molecules, which resulted in different response signals [Figure 1]a. Sensors BB, CA, and HA showed the strongest response to most of the samples. Among these three, the sensor BB exhibited the strongest signal response with 2400 above to all samples compared with other sensors, followed by CA and HA sensors with approximately about 2100 above. PE was not sensitive to all samples with a signal response of no more than 1600. [Figure 1]d indicates that the seven sensor signals of P. notoginseng samples were higher than other species in the Panax genus. In comparison, signals of P. japonicus var. major were relatively lower among the five species. The result indicated that the E-tongue might be a tool to discriminate the samples in a specific genus.
|Figure 1: The responses of the array of sensors in the electronic tongue to samples. Histogram and heatmap for five species, namely, Panax japonicus, Panax japonicus var. major, Panax zingiberensis, Panax quinquefolius, and Panax notoginseng (a and d), Panax japonicus from four areas (b and e), Panax japonicus var. major from four areas (c and f)|
Click here to view
As for P. japonicus samples [Figure 1]b from Sichuan, Shanxi, Yunnan, and Hubei Provinces in China, the sensor BB has the most forceful response to all samples (varied from 2400 to 2600). Moreover, the sensors JE and JB reached the lowest signals compared with other sensors (1400 and 1700, respectively). [Figure 1]e shows that the six sensor signals of the P. japonicus sample from Shanxi was much lower than the others, except the HC sensor signal, which was higher than those from Hubei and Yunnan. Five sensor signal responses from Hubei Province were slightly higher compared with the others, except for the JE and HA sensors, which were lower. The other sensors changed mostly in different area samples.
As for P. japonicus var. major samples [Figure 1]c from Sichuan, Yunnan, Gansu, and Tibet, the signals of sensors BB, CA, and HA were most sensitive to samples, reaching approximately 2600, 2200, and 2100, respectively. The signals of sensors GA, JB, and JE were lower than the other sensors, among which JE is the lowest with about 1400. The signals of the remaining sensors were different for samples from varying geographies. The response of seven sensors to P. japonicus var. major samples from different places seemed to be equal with little difference. [Figure 1]f presents that the signals of plants from the Tibet Province were higher than those from the other three producing areas. This finding indicated that E-tongue measurements could be useful for identifying samples from different areas for a certain species.
Principal component analysis
In the preliminary data analysis, PCA was used to study any possible clustering of samples according to the five species' signal response acquired by the E-tongue. The application of PCA to the total dataset of all features extracted from the sensor array in the presence of five species shows a separation between species with the three components (PC) explaining 91.44% of the total variance [Figure 2]a. PC1 was the component that underwent the main variations with more than 70% of the information. PC2 and PC3 were 15% and 6%, respectively. Some species of Panax genus were differentiated clearly from the others, whereas others were not so distinctive. For instance, it is straightforward to discriminate P. japonicus var. major, P. quinquefolius, and P. zingiberensis from P. japonicus and P. notoginseng. However, it is difficult to discriminate P. japonicus and P. notoginseng, because they are close to each other in the PCA plot.
|Figure 2: 3D principal component analysis plot of electronic tongue sensor signal to samples. (a) Panax japonicus, Panax japonicus var. major, Panax zingiberensis, Panax quinquefolius, and Panax notoginseng; (b) Panax japonicus from Hubei, Sichuan, Shanxi, and Yunnan Provinces. (c) Panax japonicus var. major from Sichuan, Gansu, Yunnan, and Tibet|
Click here to view
[Figure 2]b presents a 3D PCA score plot of P. japonicus from different locations. Furthermore, PC1, PC2, and PC3 were 75.3%, 13.3%, and 5.8%, respectively. In general, most samples can be roughly classified, whereas it is difficult to classify the samples from Yunnan Province, because one of them was so close to samples from Hubei Province. Samples from Sichuan and Shanxi Provinces were identified among others, because they were far from other samples and separated from each other.
[Figure 2]c shows a 3D PCA score plot results of P. japonicus var. major samples from different places in China, and several trends were observed. The accumulated explained variance was 94.6%, which was distributed in 79.0% (PC1), 10.8% (PC2), and 5.9% (PC3). Twelve samples of P. japonicus var. major of can be classified in general. Moreover, P. japonicus var. major samples from Sichuan and Gansu were discriminated clearly among samples from other locations. Similar samples appeared in the same location as in the graph. Thus, the samples from Xizang and Yunnan were relatively close to each other, which meant that their chemical compounds may be similar.
Hierarchical cluster analysis
An unsupervised HCA method was applied to study the signal response. The dendrogram obtained is shown in [Figure 3]. It is of great importance to highlight that the branch length in the dendrogram is correlated with the distances between the various clusters. Hence, it is a method to estimate their similarity. In the present experiment, sample similarities were calculated based on the Euclidean distance and the Ward's method. Therefore, two similar clusters depicted by two connected small branches shared a high similarity index. As shown in [Figure 3], each sample in the Panax genus of the same grade level was tightly clustered but sufficiently distinct from each other in the dendrogram with no misclassifications. Using a similarity index of approximately 300 in the Panax genus, five clusters were visualized.
|Figure 3: Hierarchical cluster analysis of Panax genus samples. (a) Panax japonicus, Panax japonicus var. major, Panax zingiberensis, Panax quinquefolius, and Panax notoginseng; (b) Panax japonicus from Hubei, Sichuan, Shanxi, and Yunnan Provinces. (c) Panax japonicus var. major from Sichuan, Gansu, Yunnan, and Tibet|
Click here to view
As for P. japonicus from different geographic origins [Figure 3]b, the samples from Sichuan Province were clustered together, as well as the Hubei Province samples. However, one P. japonicus sample from Yunnan Province shared high similarity with that in the Hubei Province, and one sample produced in Shanxi was close to samples cultivated in Sichuan Province. This finding indicated that the HCA method can identify samples from different areas to some degree.
[Figure 3]c shows the HCA for P. japonicus var. major from different producing areas in China. Samples cultivated in the same place can be classified without misclassification. This separation was in good agreement with the PCA results, in which all samples were distinguished according to their geographic origins.
Linear discriminant analysis
LDA is a supervised classification method and has been used to develop the classifier model. Compared with the PCA and HCA models, the LDA model shows clearer discrimination among five species in the Panax genus, as shown in [Figure 4]a. The explained variances by each discriminant function (DF) were 88.2.0% (DF1) and 9.3% (DF2) for Panax genus. Each species sample can be distinctly classified with the others. As a result, the LDA model is a superior method to discriminate the five species of Panax Linn. from all over the world. [Figure 4]b presents the LDA result of P. japonicus samples from different geographical origins. The 12 samples can be classified based on their producing areas with 70.9% in DF1 and 19.8% in DF2. [Figure 4]c presents the LDA result of P. japonicus var. major samples from different producing areas. The 12 samples can be classified into four groups according to their producing areas with 71.6% in DF1 and 21.5% in DF2.
|Figure 4: Discrimination of linear discriminant analysis plots for Panax genus. (a) Panax japonicus, Panax japonicus var. major, Panax zingiberensis, Panax quinquefolius, and Panax notoginseng; (b) Panax japonicus from Hubei, Sichuan, Shanxi, and Yunnan Provinces. (c) Panax japonicus var. major from Sichuan, Gansu, Yunnan, and Tibet. (d) The prediction of Panax zingiberensis and Panax quinquefolius using the established model of five species|
Click here to view
The performance of the LDA model was evaluated. [Table 2] shows the confusion matrix of the LDA classifier. Rows indicate actual categories, and columns are the predicted ones. The ideal situation occurs when all the samples of every species end up on the diagonal cells of the matrix. Therefore, 100% accuracy in the recognition of the Panax genus collected from five species all over the world was achieved.
|Table 2: Confusion matrix for linear discriminant analysis prediction method of Panax genus from four producing areas|
Click here to view
The LDA model can classify the sample, predict the unknown sample, and predict which classification group it belongs to. Thus, we used the LDA model to predict P. japonicus sample from Yunnan in China (predicted PZ) and P. quinquefolius sample from Canada (predicted PQ) and to calculate which group they belong to based on established DFs. The result is shown in [Figure 4]d. Predicted PZ was classified to P. zingiberensis, and predicted PQ was classified to P. quinquefolius. The LDA model can predict the two samples without any misclassification.
| Discussion|| |
In this study, E-tongue technology combined with chemometrics was first used to assess the taste of five species of the Panax genus plant from China, the USA, and Myanmar and four different geographical origins of P. japonicus and P. japonicus var. major. Among the three models, the LDA model showed the clearest discrimination. It demonstrated that the voltammetric electronic tongue system, formed by seven electrodes, can differentiate among the species of the Panax genus according to their botanical and geographical origins. The operations of data acquisition and processing in the E-tongue are simpler and more convenient than many traditional classified approaches. Therefore, this simple method based on an E-tongue could be useful as a replacement tool for the traditional analytical methods used in classification and prediction. Their use as a low-cost, continuous, nondestructive quality control system would result in significant savings in terms of labor and time. Researchers need to focus on the connection between the flavor and chemical constituents of Panax genus samples and correlating E-tongue signals with human perceptions of taste.
This work was supported by the National Natural Science Foundation of China (NO.81473315), National Science and Technology Fundamental Resources Investigation Program of China (2018FY100701), Sichuan Province Science and Technology Plan Project (2018JZ0028), the Open Research Fund of Chengdu University of Traditional Chinese Medicine Key Laboratory of Systematic Research of Distinctive Chinese Medicine Resources in Southwest China (003109034001), and CAMS Innovation Fund for Medical Sciences (no: 2016-I2M-3-015).
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Qi LW, Wang CZ, Yuan CS. Isolation and analysis of ginseng: Advances and challenges. Nat Prod Rep 2011;28:467-95.
Wang T, Guo R, Zhou G, Zhou X, Kou Z, Sui F, et al
. Traditional uses, botany, phytochemistry, pharmacology and toxicology of Panax notoginseng (Burk.) F.H. Chen: A review. J Ethnopharmacol 2016;188:234-58.
Wang Y, Choi HK, Brinckmann JA, Jiang X, Huang L. Chemical analysis of Panax quinquefolius
(North American ginseng): A review. J Chromatogr A 2015;1426:1-5.
Xu LL, Han T, Wu JZ, Zhang QY, Zhang H, Huang BK, et al
. Comparative research of chemical constituents, antifungal and antitumor properties of ether extracts of Panax ginseng and its endophytic fungus. Phytomedicine 2009;16:609-16.
Pace R, Martinelli EM, Sardone N, D E Combarieu E. Metabolomic evaluation of ginsenosides distribution in Panax genus (Panax ginseng and Panax quinquefolius
) using multivariate statistical analysis. Fitoterapia 2015;101:80-91.
Shi XJ, Yang WZ, Qiu S, Yao CL, Shen Y, Pan HQ, et al
. An in-source multiple collision-neutral loss filtering based nontargeted metabolomics approach for the comprehensive analysis of malonyl-ginsenosides from Panax ginseng, P. quinquefolius
, and P. notoginseng
. Anal Chim Acta 2017;952:59-70.
Zhang S, Wang R, Zeng W, Zhu W, Zhang X, Wu C, et al
. Resource investigation of traditional medicinal plant Panax japonicus
(T.Nees) C.A. Mey and its varieties in China. J Ethnopharmacol 2015;166:79-85.
Liu Y, Zhao J, Chen Y, Li W, Li B, Jian Y, et al
. Polyacetylenic oleanane-type triterpene saponins from the roots of Panax japonicus
. J Nat Prod 2016;79:3079-85.
He H, Xu J, Xu Y, Zhang C, Wang H, He Y, et al
. Cardioprotective effects of saponins from Panax japonicus
on acute myocardial ischemia against oxidative stress-triggered damage and cardiac cell death in rats. J Ethnopharmacol 2012;140:73-82.
Chen W, Balan P, Popovich DG. Comparison of ginsenoside components of various tissues of new Zealand forest-grown Asian ginseng (Panax Ginseng) and American Ginseng (Panax quinquefolium L.). Biomolecules 2020;10:372.
Li CT, Wang HB, Xu BJ. A comparative study on anticoagulant activities of three Chinese herbal medicines from the genus Panax and anticoagulant activities of ginsenosides Rg1 and Rg2. Pharm Biol 2013;51:1077-80.
Sun S, Qi LW, Du GJ, Mehendale SR, Wang CZ, Yuan CS. Red notoginseng: Higher ginsenoside content and stronger anticancer potential than Asian and American ginseng. Food Chem 2011;125:1299-305.
Du Z, Li J, Zhang X, Pei J, Huang L. An Integrated LC-MS-Based strategy for the quality assessment and discrimination of three panax species. Molecules 2018;23:2988.
Chan TW, But PP, Cheng SW, Kwok IM, Lau FW, Xu HX. Differentiation and authentication of Panax ginseng, Panax quinquefolius, and ginseng products by using HPLC/MS. Anal Chem 2000;72:1281-7.
Zheng S, Ren W, Huang L. Geoherbalism evaluation of Radix Angelica sinensis based on electronic nose. J Pharm Biomed Anal 2015;105:101-6.
Si R, Han Y, Wu D, Qiao F, Bai L, Wang Z, et al
. Ionic liquid organic functionalized ordered mesoporous silica integrated dispersive solid phase extraction for determination of plant growth regulators in fresh Panax ginseng. Talanta 2020;207:120247.
Yang H, Lee DY, Kang KB, Kim JY, Kim SO, Yoo YH, et al
. Identification of ginsenoside markers from dry purified extract of Panax ginseng by a dereplication approach and UPLC-QTOF/MS analysis. J Pharm Biomed Anal 2015;109:91-104.
Zhang L, Shen H, Xu J, Xu JD, Li ZL, Wu J, et al
. UPLC-QTOF-MS/MS-guided isolation and purification of sulfur-containing derivatives from sulfur-fumigated edible herbs, a case study on ginseng. Food Chem 2018;246:202-10.
Liu Y, Wang X, Wang L, Chen X, Pang X, Han J. A nucleotide signature for the identification of American ginseng and its products. Front Plant Sci 2016;7:319.
Xian-Qiao H, Zhen-Ling G, Zhi-Wei Z. Volatile compounds, affecting factors and evaluation methods for rice aroma: A review. Trends Food Sci Technol 2020, 97:136-46.
Li Z, Paul R, Ba Tis T, Saville AC, Hansel JC, Yu T, et al
. Non-invasive plant disease diagnostics enabled by smartphone-based fingerprinting of leaf volatiles. Nat Plants 2019;5:856-66.
El Alami El Hassani N, Tahri K, Llobet E, Bouchikhi B, Errachid A, Zine N, et al
. Emerging approach for analytical characterization and geographical classification of Moroccan and French honeys by means of a voltammetric electronic tongue. Food Chem 2018;243:36-42.
Podrażka M, Bączyńska E, Kundys M, Jeleń PS, Witkowska Nery E. Electronic tongue a tool for all tastes? Biosensors (Basel) 2017;8:3.
Tahara Y, Toko K. Electronic tongues A review. IEEE Sensors J 2013;13:3001-11.
Wang H, Sun H. Potential use of electronic tongue coupled with chemometrics analysis for early detection of the spoilage of Zygosaccharomyces rouxii in apple juice. Food Chem 2019;290:152-8.
Wasilewski T, Migoń D, Gębicki J, Kamysz W. Critical review of electronic nose and tongue instruments prospects in pharmaceutical analysis. Anal Chim Acta 2019;1077:14-29.
Xu M, Yang SL, Peng W, Liu YJ, Xie DS, Li XY, et al
. A novel method for the discrimination of semen arecae and its processed products by using computer vision, electronic nose, and electronic tongue. Evid Based Complement Alternat Med 2015;2015:753942.
Zou G, Xiao Y, Wang M, Zhang H. Detection of bitterness and astringency of green tea with different taste by electronic nose and tongue. PLoS One 2018;13:e0206517.
Margaritis A, Soenen H, Fransen E, Pipintakos G, Jacobs G, Blom J, et al
. Identification of ageing state clusters of reclaimed asphalt binders using principal component analysis (PCA) and hierarchical cluster analysis (HCA) based on chemo-rheological parameters. Construct Build Mater 2020;244:118276.
Bougrini M, Tahri K, Saidi T, El Hassani NA, Bouchikhi B, El Bari N. Classification of honey according to geographical and botanical origins and detection of its adulteration using voltammetric electronic tongue. Food Analytical Methods 2016;9:2161-73.
Ding J, Gu C, Huang L, Tan R. Discrimination and geographical origin prediction of Cynomorium songaricum
rupr. from different growing areas in china by an electronic tongue. J Anal Methods Chem2018;2018:1-6.
Granato D, Santos JS, Escher GB, Ferreira BL, Maggio RM. Use of principal component analysis (PCA) and hierarchical cluster analysis (HCA) for multivariate association between bioactive compounds and functional properties in foods: A critical perspective. Trends Food Sci Technol 2018;72:83-90.
Berna R, Mitra N, Hoffstad O, Wan J, Margolis DJ. Identifying phenotypes of atopic dermatitis in a longitudinal United States cohort using unbiased statistical clustering. J Invest Dermatol 2020;140:477-9.
Joseph V, Kopke LJ , Michael DE. Applying LDA-based pattern recognition to predict isometric shoulder and elbow torque generation in individuals with chronic stroke with moderate to severe motor impairment. J Neuroeng Rehabil 2019;16:35.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2]