Spotting Frozen Curd in PDO Buffalo Mozzarella Cheese Through Insights on Its Supramolecular Structure Acquired by 1H TD-NMR Relaxation Experiments
| Autor: | Davide Rabiti, Gianfranco Picone, Pasquale Ferranti, Alessandra Aiello, Francesco Capozzi, Raffaele Romano, Carlo Mengucci, Eleonora Urbinati |
|---|---|
| Přispěvatelé: | Mengucci, Carlo, Rabiti, Davide, Urbinati, Eleonora, Picone, Gianfranco, Romano, Raffaele, Aiello, Alessandra, Ferranti, Pasquale, Capozzi, Francesco |
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
relaxometry
Supramolecular chemistry 01 natural sciences lcsh:Technology Matrix (chemical analysis) Chemometrics lcsh:Chemistry food structure General Materials Science Food structure Instrumentation lcsh:QH301-705.5 Mathematics Fluid Flow and Transfer Processes 010405 organic chemistry lcsh:T Process Chemistry and Technology 010401 analytical chemistry Relaxation (NMR) General Engineering Spotting Quadratic classifier chemometrics lcsh:QC1-999 0104 chemical sciences Computer Science Applications machine learning lcsh:Biology (General) lcsh:QD1-999 lcsh:TA1-2040 Biological system lcsh:Engineering (General). Civil engineering (General) Mozzarella cheese lcsh:Physics chemometric |
| Zdroj: | Applied Sciences, Vol 11, Iss 1466, p 1466 (2021) Applied Sciences Volume 11 Issue 4 |
| Popis: | The addition of frozen curd (FC) during the production process of “Mozzarella di Bufala Campana”, an Italian cheese with Protected Designation of Origin (PDO), is a common fraud not involving modifications of the chemical composition in the final product. Its detection cannot thus be easily obtained by common analytical methods, which are targeted at changes in concentrations of diagnostic chemical species. In this work, the possibility of spotting this fraud by focusing on the modifications of the supramolecular structure of the food matrix, detected by time domain nuclear magnetic resonance (TD-NMR) experiments, was investigated. Cheese samples were manufactured in triplicate, according to the PDO disciplinary of production, except for using variable amounts of FC (i.e., 0, 15, 30, and 50% w/w). Relaxation data were analysed through different approaches: (i) Discrete multi-exponential fitting, (ii) continuous Laplace inverse fitting, and (iii) chemometrics approach. The strategy that lead to best detection results was the chemometrics analysis of raw Carr-Purcell-Meiboom-Gill (CPMG) decays, allowing to discriminate between compliant and adulterated samples, with as low as 15% of FC addition. The strategy is based on the use of machine learning for projection on latent structures of raw CPMG data and classification tasks for fraud detection, using quadratic discriminant analysis. By coupling TD-NMR raw decays with machine learning, this work opens the way to set up a system for detecting common food frauds modifying the matrix structure, for which no official authentication methods are yet available. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|