Premium
A Novel Approach in Herbal Quality Control Using Hyperspectral Imaging: Discriminating Between Sceletium tortuosum and Sceletium crassicaule
Author(s) -
Shikanga Emmanuel Amukohe,
Viljoen Alvaro M.,
Vermaak Ilze,
Combrinck Sandra
Publication year - 2013
Publication title -
phytochemical analysis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.574
H-Index - 72
eISSN - 1099-1565
pISSN - 0958-0344
DOI - 10.1002/pca.2431
Subject(s) - hyperspectral imaging , chemometrics , principal component analysis , partial least squares regression , chemistry , chromatography , linear discriminant analysis , chemical imaging , pattern recognition (psychology) , artificial intelligence , mathematics , statistics , computer science
Sceletium tortuosum is the most sought after species of the genus Sceletium and is commonly included in commercial products for the treatment of psychiatric conditions and neurodegenerative diseases. However, this species exhibits several morphological and phytochemical similarities to S. crassicaule . Objectives The aim of this investigation was to use ultrahigh‐performance liquid chromatography (UPLC) and hyperspectral imaging, in combination with chemometrics, to distinguish between S. tortuosum and S. crassicaule , and to accurately predict the identity of specimens of both species. Methods Chromatographic profiles of S. tortuosum and S. crassicaule specimens were obtained using UPLC with photodiode array detection. A SisuChema near infrared hyperspectral imaging camera was used for acquiring images of the specimens and the data was processed using chemometric computations. Results Chromatographic data for the specimens revealed that both species produce the psychoactive alkaloids that are used as quality control biomarkers. Principal component analysis of the hyperspectral image of reference specimens for the two species yielded two distinct clusters, the one representing S. tortuosum and the other representing S. crassicaule . A partial least squares discriminant analysis model correctly predicted the identity of an external dataset consisting of S. tortuosum or S. crassicaule samples with high accuracy (>94%). Conclusions A combination of hyperspectral imaging and chemometrics offers several advantages over conventional chromatographic profiling when used to distinguish S. tortuosum from S. crassicaule . In addition, the constructed chemometric model can reliably predict the identity of samples of both species from an external dataset. Copyright © 2013 John Wiley & Sons, Ltd.