Electronic nose and isotope ratio mass spectrometry in combination with chemometrics for the characterization of the geographical origin of Italian sweet cherries

•Geographic origin of sweet cherries was discriminated.•Different strategies for electronic nose variable reduction were tested.•Isotope ratio mass spectrometry gave the best model performances.•External validation was carried out on the statistical models. Sweet cherries from two Italian regions, A...

Full description

Saved in:
Bibliographic Details
Published in:Food chemistry 2015-03, Vol.170, p.90-96
Main Authors: Longobardi, F., Casiello, G., Ventrella, A., Mazzilli, V., Nardelli, A., Sacco, D., Catucci, L., Agostiano, A.
Format: Article
Language:English
Subjects:
Chemometrics
Cherries
Classification
Electronic Nose
Electronic noses
Geographic origin
Geography
Isotope ratio mass spectrometry
Isotope ratios
Isotopes - analysis
Italy
Mass spectrometry
Mass Spectrometry - methods
Mathematical models
Origins
Prunus
Prunus avium - chemistry
Spectrum Analysis - methods
Sweet cherry
Sweets
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Geographic origin of sweet cherries was discriminated.•Different strategies for electronic nose variable reduction were tested.•Isotope ratio mass spectrometry gave the best model performances.•External validation was carried out on the statistical models. Sweet cherries from two Italian regions, Apulia and Emilia Romagna, were analysed using electronic nose (EN) and isotope ratio mass spectrometry (IRMS), with the aim of distinguishing them according to their geographic origin. The data were elaborated by statistical techniques, examining the EN and IRMS datasets both separately and in combination. Preliminary exploratory overviews were performed and then linear discriminant analyses (LDA) were used for classification. Regarding EN, different approaches for variable selection were tested, and the most suitable strategies were highlighted. The LDA classification results were expressed in terms of recognition and prediction abilities and it was found that both EN and IRMS performed well, with IRMS showing better cross-validated prediction ability (91.0%); the EN–IRMS combination gave slightly better results (92.3%). In order to validate the final results, the models were tested using an external set of samples with excellent results.
ISSN:0308-8146
1873-7072
DOI:10.1016/j.foodchem.2014.08.057