Geographical discrimination of Asian red pepper powders using 1H NMR spectroscopy and deep learning-based convolution neural networks

This study investigated an innovative approach to discriminate the geographical origins of Asian red pepper powders by analyzing one-dimensional 1H NMR spectra through a deep learning-based convolution neural network (CNN). 1H NMR spectra were collected from 300 samples originating from China, Korea, and Vietnam and used as input data. Principal component analysis - linear discriminant analysis and support vector machine models were employed for comparison. Bayesian optimization was used for hyperparameter optimization, and cross-validation was performed to prevent overfitting. As a result, all three models discriminated the origins of the test samples with over 95 % accuracy. Specifically, the CNN models achieved a 100 % accuracy rate. Gradient-weighted class activation mapping analysis verified that the CNN models recognized the origins of the samples based on variations in metabolite distributions. This research demonstrated the potential of deep learning-based classification of 1H NMR spectra as an accurate and reliable approach for determining the geographical origins of various foods.

Isotopic characteristics (δ13C, δ15N, and δ18O) of honey from Bangladesh retail markets: Investigating sugar manipulation, botanical and geographical authentication

The study analyzed stable isotope composition (carbon, nitrogen, oxygen) of Bangladesh origin monofloral and multifloral honey for the first time to identify the C-4plant sugar adulteration, botanical and geographical differentiation. The C-4 sugar content (11.90 to 86.61%) using carbon isotope values of whole honey and protein extractidentified18% adulterated honeys.Whereas 82% honey have been detected as authentic with the δ13C ranges from -24.86 to -29.26‰ and the C-4 sugar values ranges from -6.75 to 5.67%. The chemometric approach using carbon and nitrogen isotopesvalues revealed discrimination among floral types of Bangladeshi authentic honeys indicating the influence of botanical source on isotopic composition. Canonical discriminant analysis (CDA) of C and N isotopes showed geographic distinction among Bangladeshi and overseas origin honeys. It is notable that the black seed honeys were correctly classified with the CDA analysis representing this type as unique Bangladesh brand as New Zealand manuka honey.

Discriminating geographical origins and determining active substances of water caltrop shells through near-infrared spectroscopy and chemometrics

Near-infrared spectroscopy (NIRS) combined with chemometric methods were used to discriminate the geographical origins of the water caltrop shells from different regions of China. Two active substances, the total phenolic content (TPC) and total flavonoid content (TFC) in the water caltrop shells were determined through the technique as well. Principal component analysis (PCA) combined with linear discriminant analysis (LDA) was adopted to build the geographical discriminant model. Quantitative analysis models of TPC and TFC were built using partial least squares (PLS) regression. 1st derivative and randomization test (RT) methods were used to optimize the quantitative analysis models. It was found that the geographical discriminant model can correctly recognize the water caltrop shells from different regions of China with a total accuracy of 93.33%. The values of TPC and TFC obtained by the optimized models and the standard method are close. The coefficient of determination (R2) and the ratio of prediction to deviation for the two substances were 0.91, 0.89 and 3.02, 3.02, respectively. The results demonstrated the feasibility of NIRS combined with chemometric methods for the geographical discrimination of water caltrop shells and the quantitative analysis of TPC and TFC in water caltrop shells.

Geographical discrimination of Flos Trollii by GC-MS and UHPLC-HRMS-based untargeted metabolomics combined with chemometrics

For centuries, Flos Trollii has been consumed as functional tea and a folk medicine in China's north and northwest zones. The quality of Flos Trollii highly depends on the producing zones. Unfortunately, few studies have been reported on the geographical discrimination of Flos Trollii. This work comprehensively investigated Flos Trollii compounds with an integration strategy combining gas chromatography-mass spectrometry (GC-MS) and ultrahigh-performance liquid chromatography-high resolution mass spectrometry (UHPLC-HRMS) with chemometrics to explore the differences between Flos Trollii obtained from various origins of China. About 71 volatile and 22 involatile markers were identified with GC-MS and UHPLC-HRMS, respectively. Geographical discrimination models were synthetically investigated based on the identified markers. The results indicated that the UHPLC-HRMS coupled with the fisher discrimination model provided the best prediction capability (>97%). This study provides a new solution for Flos Trollii discrimination.

Geographical discrimination of dried chili peppers using femtosecond laser ablation-inductively coupled plasma-mass spectrometry (fsLA-ICP-MS)

This study presents a method for discriminating the geographical origin of dried chili peppers using femtosecond laser ablation-inductively coupled plasma-mass spectrometry (fsLA-ICP-MS) and multivariate analysis, such as orthogonal partial least squares discriminant analysis (OPLS-DA), heatmap analysis, and canonical discriminant analysis (CDA). Herein, 102 samples were analyzed for the content of 33 elements using optimized conditions of 200 Hz (repetition rate), 50 μm (spot size), and 90% (energy). Significant differences in count per second (cps) values of the elements were observed between domestic and imported peppers, with variations of up to 5.66 times (¹³³Cs). The OPLS-DA model accuracy achieved an R² of 0.811 and a Q² of 0.733 for distinguishing dried chili peppers of different geographical origins. The variable importance in projection (VIP) and s-plot identified elements 10 and 3 as key to the OPLS-DA model, and in the heatmap, six elements were estimated to be significant in discriminating between domestic and imported samples. Furthermore, CDA showed a high accuracy of 99.02%. This method can ensure food safety for consumers, and accurately determine the geographic origin of agricultural products.

Geographical origin traceability of Keemun black tea based on its non-volatile composition combined with chemometrics

BACKGROUND

Non-volatile compounds play a key role in the quality and price of Keemun black tea (KBT). The non-volatile compounds in KBT samples from different producing areas normally vary greatly. The development of rapid methods for tracing the geographical origin of KBT is useful. In this study, we develop models for the discrimination of KBT's geographical origin based on non-volatile compounds.

RESULTS

Seventy-two KBT samples were collected from five towns in Anhui province to determine 13 KBT compounds by high-performance liquid chromatography (HPLC). Analysis of variance showed that the content of 13 compounds in KBT indicated significant differences (P < 0.05) among five towns. Three multivariate statistical models including principal component analysis (PCA), soft independent modeling of class analogy (SIMCA), and linear discriminant analysis (LDA) were built to discriminate origin. Principal component analysis effectively extracted three principal components, namely theaflavins, galloylated catechins, and simple catechins. The high sensitivity (64.5%-99.2%) was achieved of SIMCA model. To establish the discriminant functions, six variables (gallic acid, (+)-catechin, (-)-epigallocatechin gallate, theaflavin-3-gallate, theaflavin-3,3'-di-gallate, and total theaflavins) were chosen from 13 variables, and LDA was applied. This gave a satisfactory overall correct classification rate (94.4%) and cross-validation rate (88.9%) for KBT samples.

CONCLUSION

The results showed that HPLC analysis together with chemometrics is a reliable approach for tracing KBT and guaranteeing its authenticity. © 2019 Society of Chemical Industry.

Data fusion for food authentication. Combining rare earth elements and trace metals to discriminate "Fava Santorinis" from other yellow split peas using chemometric tools

"Fava Santorinis", is a protected designation of origin (PDO) yellow split pea species growing only in the island of Santorini in Greece. Due to its nutritional quality and taste, it has gained a high monetary value. Thus, it is prone to adulteration with other yellow split peas. In order to discriminate "Fava Santorinis" from other yellow split peas, four classification methods utilising rare earth elements (REEs) measured through inductively coupled plasma-mass spectrometry (ICP-MS) are studied. The four classification processes are orthogonal projection analysis (OPA), Mahalanobis distance (MD), partial least squares discriminant analysis (PLS-DA) and k nearest neighbours (KNN). Since it is known that trace elements are often useful to determine geographical origin of food products, we further quantitated for trace elements using ICP-MS. Presented in this paper are results using the four classification processes based on the fusion of the REEs data with the trace element data. Overall, the OPA method was found to perform best with up to 100% accuracy using the fused data.