mid-infrared spectroscopy and short wave infrared hyperspectral imaging—a novel approach in the...

1

2

3

4

5 Q1

67

8

9

1011

12

13

14

15

16

17

18

19

20

21

22

23

Phytochemistry Letters xxx (2013) xxx–xxx

Q2

G Model

PHYTOL 629 1–7

Mid-infrared spectroscopy and short wave infrared hyperspectralimaging—A novel approach in the qualitative assessment ofHarpagophytum procumbens and H. zeyheri (Devil’s Claw)

Nontobeko Mncwangi a, Ilze Vermaak a, Alvaro M. Viljoen a,b,*a Department of Pharmaceutical Sciences, Faculty of Science, Tshwane University of Technology, Private Bag X680, Pretoria 0001, South Africab Department of Pharmaceutics and Industrial, Faculty of Pharmacy, King Abdulaziz University, Jeddah 21589, Saudi Arabia

A R T I C L E I N F O

Article history:

Received 12 September 2013

Received in revised form 13 November 2013

Accepted 15 November 2013

Available online xxx

Keywords:

Chemometrics

Devil’s Claw

Harpagophytum

Hyperspectral imaging

Spectroscopy

Quality control

A B S T R A C T

Harpagophytum procumbens (Burch.) DC. ex Meisn. subsp. procumbens (Pedaliaceae) is an important

African medicinal plant growing in the Kalahari region of southern Africa. This species, together with its

close taxonomic ally Harpagophytum zeyheri are collectively referred to as Devil’s Claw and are used

interchangeably for the treatment of inflammation-related disorders. Although the two taxa are

botanically and chemically similar, H. zeyheri contains lower levels of harpagoside and these two species

have not been proven to exhibit equipotent pharmacological activity. Due to these taxonomic

similarities, effective quality control methods are required to distinguish between the two species.

Differentiation between the two species was achieved using single point mid-infrared spectroscopy in

combination with chemometric data analysis. The orthogonal projections to latent structures

discriminant analysis (OPLS-DA) model had good predictive ability, as illustrated by the model

statistics: R2X (cum predictive + orthogonal) = 0.86 and Q2 (cum) = 0.63. Short wave infrared (SWIR)

hyperspectral imaging could distinguish between the two species with acceptable model statistics: R2X

and R2Y of 0.99 and 0.78, respectively. This study demonstrated that both MIR single point spectroscopy

and SWIR hyperspectral imaging coupled with chemometric modelling is a reliable and rapid method to

determine the authenticity of Harpagophytum spp.

� 2013 Published by Elsevier B.V. on behalf of Phytochemical Society of Europe.

Contents lists available at ScienceDirect

Phytochemistry Letters

jo u rn al h om ep ag e: ww w.els evier .c o m/lo c ate /p hyt ol

2425262728293031323334353637

1. Introduction

Harpagophytum procumbens (Burch.) DC. ex Meisn. subsp.procumbens (Pedaliaceae) is an important African medicinal plantfound in the Kalahari region of southern Africa which includesNamibia, South Africa, Botswana, Angola, Zimbabwe, Zambia andMozambique. According to ethnobotanical information, secondaryroot tubers are used to treat painful inflammation of the joints andmuscles in rheumatism and arthritis, back pain, and as a tonic forgastro-enterological disturbances (Mncwangi et al., 2012; van Wykand Gericke, 2000). A topical ointment prepared with animal fat orpetroleum is traditionally applied to treat sores, ulcers and boils(van Wyk and Gericke, 2000). Other traditional uses of H.

procumbens include the treatment of fever, diabetes, diarrhoea
3839404142434445
Abbreviations: ATR, attenuated total reflectance; MIR, mid-infrared; OPLS-DA,

orthogonal projections to latent structures discriminant analysis; PCA, principal

component analysis; SWIR, short wave infrared; SNV, standard normal variate.

* Corresponding author at: Department of Pharmaceutical Sciences, Faculty of

Science, Tshwane University of Technology, Private Bag X680, Pretoria 0001, South

Africa. Tel.: +27 12 382 6373; fax: +27 12 382 6243.

E-mail addresses: [email protected], [email protected] (A.M. Viljoen).

Please cite this article in press as: Mncwangi, N., et al., Mid-infrarednovel approach in the qualitative assessment of Harpagophytum pro

http://dx.doi.org/10.1016/j.phytol.2013.11.007

1874-3900/$ – see front matter � 2013 Published by Elsevier B.V. on behalf of Phytoc


and blood disease (Stewart and Cole, 2005). Studies haveconfirmed that extracts of the secondary tubers of H. procumbens

are effective in the treatment of degenerative rheumatoid arthritis,osteoarthritis, tendonitis, kidney inflammation, heart disease,dyspepsia and loss of appetite (Stewart and Cole, 2005; Viljoenet al., 2012).

H. procumbens and the closely related Harpagophytum zeyheri

are collectively known as Devil’s Claw and are used interchange-ably (Kemper, 1999). The British and European Pharmacopoeiasmake reference to both H. procumbens and H. zeyheri but H.

procumbens is the species of choice as it has been shown to containhigher amounts of the biologically active ingredient, harpagoside(Mncwangi et al., 2012). H. zeyheri has been included informulations as an adulterant where H. procumbens is used asstarting material (McGregor et al., 2005), largely due to thedeclining natural populations of H. procumbens. H. procumbens isextensively commercialised which creates a conducive environ-ment for adulteration which may have negative consequences forboth the trader and the consumer (Stewart and Cole, 2005).

Phytochemical studies have confirmed the presence of iridoidglycosides such as harpagoside, procumbide and harpagide,phenylethanoid glycosides such as acteoside and isoacteoside

spectroscopy and short wave infrared hyperspectral imaging—Acumbens and H. zeyheri (Devil’s Claw). Phytochem. Lett. (2013),

hemical Society of Europe.


mailto:[email protected]

mailto:[email protected]


http://www.sciencedirect.com/science/journal/18743900

http://dx.doi.org/www.elsevier.com/locate/phytol


46 an47 fla48 (G49 on50 ra51 sc52 H.

53 sp54 ba55 Di56 1657 ha

585960616263646566676869

FigQ3

pr

leg

N. Mncwangi et al. / Phytochemistry Letters xxx (2013) xxx–xxx2

G Model

PHYTOL 629 1–7

d other substances including harpagoquinones, amino acids,vonoids, phytosterols and carbohydrates in Devil’s Clawruenwald, 2002; Kikuchi et al., 1983; Kurkin, 2003). However,ly harpagoside is used as a biomarker to determine the quality ofw material and products. Various chromatographic and spectro-opic methods have been developed for the quality assurance of procumbens raw material and products. Using NIR-FT-Ramanectroscopy, Baranska et al. (2005) identified characteristic keynds in H. procumbens raw materials, ethanol extracts and tablets.agnostic key bands were observed in the frequency range of00–1700 cm�1, which is similar to the wavenumber region forrpagoside, and is assigned to –C55O, –C55C and benzene ring

. 1. (A) MIR spectra of powdered H. procumbens (blue) and H. zeyheri (red) secondary

ocumbide) isolated from H. procumbens. (C) Chemical structures of harpagoside, har

end, the reader is referred to the web version of the article.)

Please cite this article in press as: Mncwangi, N., et al., Mid-infrarednovel approach in the qualitative assessment of Harpagophytum pr


stretching vibrations. Schulz and Baranska also developed acalibration model for harpagoside using HPLC as a referencemethod. Raman mapping was used to visualise the spatialdistribution of harpagoside in different samples (Schulz andBaranska, 2007).

Quality control (QC) is a major concern in the herbal andphytomedicine industry; vibrational spectroscopy is one of themodern methods used in raw material identification and in-process monitoring (Reich, 2005). Vibrational spectroscopy offersvarious advantages over current analytical techniques; theseinclude its robustness, efficiency, non-destructiveness and cost-effectiveness. In addition, minimal or no sample preparation is

root tubers. (B) MIR spectra of three iridoid glycosides (harpagoside, harpagide and

pagide and procumbide. (For interpretation of the references to colour in this figure

spectroscopy and short wave infrared hyperspectral imaging—Aocumbens and H. zeyheri (Devil’s Claw). Phytochem. Lett. (2013),


70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100101102103104105106107108109110111112113114115116117118119120121

122123124125126127128

N. Mncwangi et al. / Phytochemistry Letters xxx (2013) xxx–xxx 3

G Model

PHYTOL 629 1–7

required (Schulz et al., 2004). The present study aimed to developvibrational spectroscopy methods to differentiate between H.

procumbens and H. zeyheri secondary root tubers which in its driedand powdered forms are indistinguishable. The feasibility of thismethod in species differentiation has been applied successfully byother researchers but has not been applied in the quality control ofDevil’s Claw.

2. Results and discussion

2.1. Mid-infrared spectroscopy fingerprints of H. procumbens and H.zeyheri

Devil’s Claw is known to comprise of phenylpropanoid glyco-sides, iridoid glycosides, acetyl phenolic glycosides, sterols andsugars such as rhamnose and rabinose (Gruenwald, 2002; Kikuchiet al., 1983; Kurkin, 2003). Characteristic spectra of H. procumbens

(blue) and H. zeyheri (red) are depicted in Fig. 1A. The MIR spectrashow characteristics key bands which can be assigned to majorcompounds in H. procumbens. Wavenumbers of major peaks wereidentified as follows: 1700–1200 cm�1: esters and phenols; 1150–950 cm�1: sugars; and 1100–1000 cm�1 is associated withalcohols (Shurvell, 2002). Fig. 1B shows the MIR spectra of threemajor iridoid glycosides from H. procumbens namely harpagoside,harpagide and procumbide and their chemical structures areshown in Fig. 1C. The highlighted region, 1700–1600 cm�1 isassigned to –C55O, –C55C and benzene ring stretching vibrationswhich is similar to the wavenumber region for harpagoside and isalso common in all three iridoid glycosides (Schulz and Baranska,2007). The sugar moieties attached to the iridoids show signals inthe region 1150–950 cm�1 as was observed in both the plantpowder and the isolated compounds.

-0.010

-0.008

-0.006

-0.004

-0.002

0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.00.0010.000-0.001-0.002-0.003-0.004-0.005-0.006-0.007

t[2]

t[4]

-0.010

-0.008

-0.006

-0.004

-0.002

0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.00-0.001-0.002-0.003-0.004-0.005-0.006-0.007

t[2]

t

A

B

Fig. 2. (A) PCA score plot of a five component model based on MIR data showing the intr

presence of geographical phytochemical variation within Devil’s Claw species.



2.2. Classification and discrimination of Harpagophytum species

2.2.1. Principal component analysis (PCA)

All 344 samples (H. zeyheri, n = 101; H. procumbens, n = 243)were included in the chemometric analysis. Baseline correctionwas applied in order to remove variation which may be due tonoise and shifts in the spectra. This was achieved through the useof chained filters (first derivative and orthogonal scatter correc-tion) and all data was mean centred. A model with five principalcomponents (PCs) was developed based on the MIR data. Themodel explained 87% of the variance R2X (cum) = 0.87 and the Q2

(cum) = 0.86 indicating that the model had good predictive ability.The clusters were not clearly demarcated with an area ofintrogression. In addition a degree of separation was noted interms of geographical location (Fig. 2A and B). van Wyk (2008)states that although the taxonomy of Harpagophytum is welldocumented, there is still a need for biosystematic studies sincethere is no clear circumscription of taxa at infraspecific level (vanWyk, 2008). PCA yielded inconclusive results as far as discrimina-tion between the two species is concerned. Although the main aimwas not to study the species complex at subspecies level; the genusis known to comprise of two species, H. zeyheri and H. procumbens

with five subspecies thus the sub-grouping may be attributed tothe differentiation of subspecies (van Wyk, 2008).

2.2.2. Orthogonal projections to latent structures-discriminant

analysis (OPLS-DA)

In order to maximise the separation of the clusters observedand to understand which variables are responsible for the class-separating information, OPLS-DA was performed. A single compo-nent, OPLS-DA model was developed for discrimination between H.

procumbens and H. zeyheri. The developed OPLS-DA model

0.0070.0060.0050.0040.00302

CasselFerrolandsGanyesaMakgabengMolopo nature reserveMoswanaSpringbokfonteinTerra fontein

0.0070.0060.0050.0040.0030.0020.0010

[4]

H. procumbensH. zeyheri

SIMCA-P+ 12 - 2013-02-04 07:41:46 (UTC+2)

ogression of Devil’s Claw species. (B) PCA score plot based on MIR data showing the



Fig. 3. (A) OPLS-DA score plot of MIR spectra for the discrimination of H. procumbens and H. zeyheri. (B) OPLS-DA S-plot showing modelled covariance vs modelled correlation,

where the positive loadings (blue) are correlated with H. procumbens and the negative loadings (red) are correlated with H. zeyheri. (C) Loadings line plot MIR data showing

wavenumbers responsible for the clustering of H. procumbens (blue) and H. zeyheri (red). (For interpretation of the references to colour in this figure legend, the reader is

referred to the web version of the article.)


G Model

PHYTOL 629 1–7

Please cite this article in press as: Mncwangi, N., et al., Mid-infrared spectroscopy and short wave infrared hyperspectral imaging—Anovel approach in the qualitative assessment of Harpagophytum procumbens and H. zeyheri (Devil’s Claw). Phytochem. Lett. (2013),http://dx.doi.org/10.1016/j.phytol.2013.11.007


129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159160161162163164165166167168169170171172173174

175

176

177178179180181182183184185186


G Model

PHYTOL 629 1–7

explained 86% of the total variation in X (R2X cum predicti-ve + orthogonal) and the Q2 (cum) was 0.63 indicating thegoodness of prediction. The first component of the OPLS-DAmodel explained 86% of variation in X (data matrix) which wasrelated to the discrimination of the two species. The goodness offit: R2Y was 70% indicating that the model was not overfitted(Fig. 3A). The S-plot (Fig. 3B) shows that the positive loadings arecorrelated to H. procumbens and the negative loadings arecorrelated to H. zeyheri. The highlighted regions are largelyresponsible for the observed separation: wavenumbers 1770–1700 cm�1 are correlated to H. procumbens spectra; wavenumbers1385–1360 cm�1 are correlated to H. zeyheri spectra (Fig. 3C).

2.2.3. Short wave infrared (SWIR) hyperspectral imaging

The removal of the background resulted in the PCA score imagedepicted in Fig. 4 showing the dried ground secondary roots only.The dataset was balanced in terms of pixel numbers for the twoclasses: the H. procumbens dataset consisted of 51.33% (13 203pixels) of the total while the H. zeyheri dataset contributed 48.59%(12 498 pixels). The PLS-DA model constructed with 3 PLS factorsshowed clear distinction between the two species in the directionof PLS factor 1 (Fig. 5). PLS factor 1 (96.40%) and PLS factor 2 (0.97%)collectively explained 97.37% of the data and the model had a goodR2X (cum) of 0.99 and R2Y (cum) of 0.78. An external validationdataset consisting of two samples, H. procumbens (sample A) and H.

zeyheri (sample B) dried secondary root powders was importedinto the PLS-DA model to predict class membership (Fig. 6). Inimage A, 98.66% of the pixels were correctly predicted as H.

procumbens. In image B, the majority of pixels (77.11%) werecorrectly predicted as H. zeyheri. The high number of pixelscorrectly predicted and the low number of unclassified pixels,

Fig. 4. The PCA score image of five H. procumbens and five H. zeyheri samples used



which is usually due to background, texture or shape effects, showsthat hyperspectral imaging is an effective quality control methodfor species authentication.

In conclusion, PCA, an unsupervised classification method,showed that there is introgression within the species complex andthere are groupings according to geographical locality. A fullunderstanding of the introgression that occurs between the twospecies and also at the subspecies level needs to be furtherexplored. However, when a supervised method, OPLS-DA, wasused, the two species were successfully separated. Although thereis evident introgression this technique may prove useful inpreventing substitution of H. procumbens with H. zeyheri. SWIRhyperspectral imaging successfully differentiated between the twospecies and could thus be used for qualitative quality control asevidenced the through authentication of samples introduced as anexternal dataset.

3. Experimental

3.1. Plant material and sample preparation

Secondary root tubers were collected from natural popula-tions in several countries (Table 1). The tubers were chopped, airdried at room temperature, milled and sieved (500 mm,Endecotts Ltd., England). The total sample was size was 344with 101 H. zeyheri and 243 H. procumbens collections . All thesamples were used for discriminant analysis through MIRspectroscopy. Twelve dried secondary root powders wereanalysed with SWIR hyperspectral imaging: ten samples wereused for model development and two samples were insertedinto the model as an external validation set.

to develop the PLS-DA model based on short wave infrared hyperspectral data.



187 3.

188

189 ac190 sp191 Gm192 so193 su194 th195 ea196 in

197

198199200201202203204205206

Fighy

(9

un

th

Fig. 5. PLS-DA score plot showing the H. procumbens ( ) and H. zeyheri ( ) clusters where PLS factor 1 (t[1]) explained 96.40% of the dataset and PLS factor 2 (t[2]) explained

0.97% of the data. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)


G Model

PHYTOL 629 1–7

2. Mid-infrared spectroscopy (MIR) measurements

The mid-infrared spectra of the dried and ground powders werequired in the range 550–4000 cm�1 on an Alpha-P Brukerectrophotometer with an ATR diamond crystal (Bruker OPTIK

bH, Ettlingen, Germany) controlled by OPUS1 (version 6.5)ftware. The powdered material was placed directly on therface of the ATR diamond crystal and spectral data obtained ine absorbance mode. A total of 32 scans were accumulated forch sample with a spectral resolution of 4 cm�1 thus accumulat-g 2436 data points.

207208209

. 6. Prediction of the external dataset imported into the PLS-DA model based on

perspectral imaging data: sample A was correctly predicted as H. procumbens

8.66%) and sample B as H. zeyheri (77.11%). H. procumbens ( ), H. zeyheri ( ),

classified ( ). (For interpretation of the references to colour in this figure legend,

e reader is referred to the web version of the article.)

Please cite this article in press as: Mncwangi, N., et al., Mid-infrarednovel approach in the qualitative assessment of Harpagophytum pr


3.3. Near infrared hyperspectral imaging camera

The SisuCHEMA short wave infrared (SWIR) hyperspectralimaging camera (Specim, Spectral Imaging Ltd., Oulu, Finland)with a spectral range of 1000–2500 nm controlled by Chemadaq(version 3.62.183.19) software was used to capture hyperspectralimages with a high magnification lens (field of view: 100 mm;spatial resolution: 150 mm). The system consisted of an imagingspectrograph coupled to a 2-D array mercury–cadmium–telluride(HgCdTe) detector. The white and black references were capturedprior the acquisition of each image to eliminate variation in sampleillumination. Image conversion to pseudo-absorbance was doneautomatically using Evince1 (version 2.4.0) multivariate analysissoftware (UmBio AB, Umea, Sweden).

Table 1Locations and number of samples (n) of Devil’s Claw collected throughout South

Africa.

Location Species Number of samples (n = 344)

Cassel 1 H. procumbens 19

Ganyesa 1 H. procumbens 54



Moswana 1 H. procumbens 25

Moswana 2 H. procumbens 23

Molopo nature reserve H. procumbens 24

Lafras H. procumbens 07

Terra Firma H. procumbens 28

Caprivi strip H. zeyheri 06

Ferrolands H. zeyheri 28

Springbokfontein H. zeyheri 21

Makgabeng H. zeyheri 46

spectroscopy and short wave infrared hyperspectral imaging—Aocumbens and H. zeyheri (Devil’s Claw). Phytochem. Lett. (2013),


210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287


G Model

PHYTOL 629 1–7

3.4. Data analysis

3.4.1. Mid-infrared (MIR) spectroscopy data

Chemometric data analysis of the single point spectral data wasperformed using SIMCA-P1 +12.0 software (Umetrics AB, Malmo,Sweden). Orthogonal projections to latent structures discriminantanalysis (OPLS-DA) was performed on the spectral data after centrescaling. Principal component analysis (PCA) was performed in orderto identify strong outliers from the scores scatter plot. Derivativesare intrinsic ways of dealing with systematic variations in thespectra and first and second derivatives are often used to reduceadditive baseline and scatter effects, respectively. Derivativesenhance properties of bands other than amplitude and intensity.First derivative emphasise band widths as most band shapes havemaxima and minima near the half-height points of the bands whilstsecond derivatives emphasise band positions and separations.

3.4.2. Short wave infrared hyperspectral image analysis

All hyperspectral data was mean centred and a principalcomponent analysis (PCA) score plot was used interactively toremove the unwanted background and edge effects (Manley et al.,2011). From the interactive PCA plot and scores image, the sampleswere assigned to two classes (H. procumbens and H. zeyheri) todevelop a PLS-DA model. Standard normal variate (SNV) pretreat-ment was applied to the data as SNV transformation reducesmultiplicative effects of scattering, particle size andmulticollinearitychanges over the whole spectral range (Zeaiter et al., 2005). Theresulting PLS-DA model with 2 classes and three PLS factors was usedto predict the species of test samples introduced into the module asan external dataset.

Acknowledgements

Dr. Z. Zietsman is thanked for assistance with the sourcing ofraw materials. The financial assistance of the National ResearchFoundation through the Thuthuka program is appreciated.



References

Baranska, M., Schulz, H., Reitzenstein, S., Uhlemann, U., Strehle, M.A., Kruger, H.,Quilitzsch, R., Foley, W., Popp, J., 2005. Vibrational spectroscopic studies toacquire a quality control method of eucalyptus essential oils. Biopolymers 78,237–248.

Gruenwald, J., 2002. Expanding the market for Devil’s Claw in Europe. In: NamibianNational Devil’s Claw Conference, Namibia.

Kemper, K.J., 1999. Devil’s Claw (Harpagophytum procumbens). Longwood HerbalTask Force.

Kikuchi, T., Matsuda, S., Kubo, Y., Namba, T., 1983. New iridoids from Harpagophy-tum procumbens D.C. Chem. Pharm. Bull. 31, 2296–2301.

Kurkin, V.A., 2003. Phenylpropanoids from medicinal plants: distribution, classifi-cation, structural analysis, and biological activity. Chem. Nat. Compd. 39,123–153.

Manley, M., du Toit, G., Geladi, P., 2011. Tracking diffusion of conditioning water insingle wheat kernels of different hardness by near infrared hyperspectralimaging. Anal. Chim. Acta 686, 64–75.

McGregor, G., Fiebich, B., Wartenberg, A., Brien, S., Lewith, G., Wegener, T., 2005.Devil’s Claw (Harpagophytum procumbens): an anti-inflammatory herb withtherapeutic potential. Phytochem. Rev. 4, 47–53.

Mncwangi, N., Chen, W., Vermaak, I., Viljoen, A.M., Gericke, N., 2012. Devil’s Claw – areview of the ethnopharmacology, phytochemistry and biological activity ofHarpagophytum procumbens. J. Ethnopharmacol. 143, 755–771.

Reich, G., 2005. Near infrared spectroscopy and imaging: basic principles andpharmaceutical applications. Adv. Drug Deliv. Rev. 57, 1109–1143.

Schulz, H., Baranska, M., 2007. Identification and quantification of valuableplant substances by IR and Raman spectroscopy. Vib. Spectrosc. 43,13–25.

Schulz, H., Baranska, M., Belz, H., Rosch, P., Strhle, M.A., Popp, J., 2004. Chemotaxo-nomic characterization of essential oil plants by vibrational spectroscopymeasurements. Vib. Spectrosc. 35, 81–86.

Shurvell, H.F., 2002. Spectra–structure correlations in the mid- and far-infrared. In:Chalmers, J.M., Griffiths, P.R. (Eds.), Handbook of Vibrational Spectroscopy. JohnWiley & Sons, New York.

Stewart, K.M., Cole, D., 2005. The commercial harvest of Devil’s Claw (Harpagophy-tum spp.) in southern Africa: the devil’s in the details. J. Ethnopharmacol. 100,225–236.

van Wyk, B.-E., 2008. Abroad review of commercially important southern Africanmedicinal plants. J. Ethnopharmacol. 119, 342–355.

van Wyk, B.-E., Gericke, N., 2000. People’s Plants: A Guide to Useful Plants ofSouthern Africa. Briza Publications, Pretoria, South Africa.

Viljoen, A., Mncwangi, N., Vermaak, I., 2012. Anti-inflammatory iridoids of botanicalorigin. Curr. Med. Chem. 19, 2104–2127.

Zeaiter, M., Roger, J.-M., Bellon-Maurel, V., 2005. Robustness of models developedby multivariate calibration. Part II: The influence of pre-processing methods.Trends Anal. Chem. 24, 437–445.



mid-infrared spectroscopy and short wave infrared hyperspectral imaging—a novel approach in the...

Documents