Rapid authentication of Chinese oolong teas using atmospheric solids analysis probe-mass spectrometry (ASAP-MS) combined with supervised pattern recognition models
| Autor: | Hui Ru Tan, Huei Hong Lee, Weibiao Zhou, Li Yan Chan, Yong-Quan Xu |
|---|---|
| Rok vydání: | 2022 |
| Předmět: |
Soft independent modelling of class analogies
business.industry Nearest neighbour Pattern recognition Mass spectrometry Linear discriminant analysis Fingerprint Principal component analysis Pattern recognition (psychology) media_common.cataloged_instance Artificial intelligence European union business Food Science Biotechnology media_common Mathematics |
| Zdroj: | Food Control. 134:108736 |
| ISSN: | 0956-7135 |
| DOI: | 10.1016/j.foodcont.2021.108736 |
| Popis: | Ambient mass spectrometry (AMS) is an emerging technique in food authenticity and traceability study due to the minimal sample preparation required and short analysis time. Herein, a non-targeted fingerprinting approach using an AMS technique, atmospheric solids analysis probe – mass spectrometry (ASAP-MS), was used to authenticate Chinese oolong teas. In the first part of the study, a total of 38 authentic samples from three main varieties – Guangdong Dancong, Taiwan Dongding, and Anxi Tieguanyin – were analysed and four discriminant analysis models were built using the fingerprint data from ASAP-MS. The principal component analysis-k nearest neighbour (PCA-kNN) model yielded the best classification outcome, where the classification accuracies of the training and validation sets were 100% and 92.6%, respectively. The second part of the study involved detecting possible adulteration of Anxi Tieguanyin, which is a high-value oolong tea under the register of protected geographical indication (PGI) of the European Union (EU). Adulteration of Anxi Tieguanyin was simulated by blending the authentic samples with 20–80% w/w of low-quality oolong teas. One-class modelling using data-driven soft independent modelling of class analogies (DD-SIMCA), with the Anxi Tieguanyin as the target class, was built using the fingerprint data of the authentic and adulterated samples. An excellent sensitivity of 100% and a high specificity of 98.1% were achieved, indicating that it is possible to detect substitution adulteration of Anxi Tieguanyin using ASAP-MS combined with one-class modelling. Overall, findings from this study exemplify the potential of ASAP-MS to be used for rapid, inexpensive, and high throughput classification of Chinese oolong tea varieties and screening for substitution adulteration of Anxi Tieguanyin oolong tea. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect