Hyperspectral imaging in combination with chemometric data analysis – a novel approach in the quality control of herbal material

The quality control of herbal material is notoriously challenging due to the complex mixture of compounds present in plants. Hyperspectral imaging (HSI) integrates conventional spectroscopy and imaging to obtain spectral and spatial information from a sample. Once the method has been developed, the visual results are rapidly obtained and easy to interpret. In this study, the use of HSI in combination with chemometric data analysis in quality control was illustrated using three examples: 1) distinguishing between the whole dried fruit of Illicium verum (Chinese star anise) and Illicium anisatum (Japanese star anise); 2) S. tetrandra ('hang fang ji') and its substitution or adulteration with Aristolochia fangchi ('guang fang ji'); 3) determining the proportion of each constituent in a tea blend consisting of Aspalathus linearis (rooibos) and Agathoshma betulina ('buchu'). Hyperspectral images were captured using a shortwave infrared pushbroom imaging system in the wavelength range 920 – 2514nm. Evince® and/or Matlab® software were used to analyse the data. For the star anise example, a classification model was developed and used to accurately predict the identity of whole dried fruit of I. anisatum and I. verum. In the 'fang ji' example, the replicates for each plant species were predicted at a value > 99% for all the samples. Artificially adulterated samples were accurately predicted from as low as 10%. In the herbal tea blend example, the classification model was applied to determine the relative proportions of each blend constituent in intact tea bags. With the increasing need to regulate herbal products and ingredients, emerging technologies are providing alternative methods that allow the holistic analysis of the samples. Hyperspectral imaging is ideally suited as a qualitative tool for the quality control of herbal raw material as it is a visual, rapid, accurate and non-destructive method with high prediction ability.