Virgin olive oil volatile fingerprint and chemometrics: Towards an instrumental screening tool to grade the sensory quality
Sensory quality, assessed following a standardized method, is one of the parameters defining the commercial category of virgin olive oil. Considering the difficulties linked to the organoleptic evaluation, especially the high number of samples to be assessed, setting up instrumental methods to suppo...
| Autores: | , , , , , , , , |
|---|---|
| Tipo de documento: | artigo |
| Estado: | Versión aceptada para publicación |
| Data de publicação: | 2020 |
| País: | España |
| Recursos: | Universidad de Barcelona |
| Repositório: | Dipòsit Digital de la UB |
| OAI Identifier: | oai:diposit.ub.edu:2445/159203 |
| Acesso em linha: | https://hdl.handle.net/2445/159203 |
| Access Level: | Acceso aberto |
| Palavra-chave: | Oli d'oliva Compostos orgànics volàtils Olive oil Volatile organic compounds |
| Resumo: | Sensory quality, assessed following a standardized method, is one of the parameters defining the commercial category of virgin olive oil. Considering the difficulties linked to the organoleptic evaluation, especially the high number of samples to be assessed, setting up instrumental methods to support sensory panels becomes a need for the olive oil sector. Volatile fingerprint by Headspace Solid Phase Microextraction-Gas Chromatography-Mass Spectrometry can be an excellent fit-for-purpose tool as the volatile fraction is responsible for virgin olive oil sensory attributes. A fingerprinting approach was applied to the volatile profile of 176 virgin olive oils pre- viously graded by six official sensory panels. The classification strategy consisted in two sequential Partial Least Square-Discriminant Analysis models built with the aligned chromatograms: the first discriminated extra virgin and non-extra virgin samples; the second classified the latter into virgin or lampante categories. Results were satisfactory in the cross-validation by leave 10%-out (97% of correct classification). For external validation, an uncertainty range was set for the prediction models to detect boundary samples, which would be further assessed by the sensory panels. By doing this, a considerable decrease of the panel workload (around 80%) was achieved, while maintaining a highly reliable classification of samples (error rate < 10%). |
|---|