U.S. flag

An official website of the United States government

Dot gov

Official websites use .gov
A .gov website belongs to an official government organization in the United States.


Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.


Main content area

Application of chemometric tools combined with instrument-agnostic GC-fingerprinting for hazelnut quality assessment

Fidel Ortega-Gavilán, Simone Squara, Chiara Cordero, Luis Cuadros-Rodríguez, Mª Gracia Bagur-González
Journal of food composition and analysis 2023 v.115 pp. 104904
Corylus avellana, chemometrics, chocolate, confectionery industry, cultivars, food composition, food quality, hazelnuts, piedmont, storage time, volatile organic compounds, Italy
The European hazelnut (Corylus avellana L.) is a tree nut that is mainly produced in Turkey, Italy and USA and used by the confectionery industry to obtain sweets and chocolate spreads. Among all the known cultivars/origins, the "Tonda Gentile Trilobata" from Piedmont (Italy) is highly appreciated due to its organoleptic properties and considered, for many applications, as a "Gold standard". Although the use of marker compounds is widely used in food quality evaluation, the fingerprinting methodology could provide additional information related to hazelnut quality by making use of non-explicit information embedded in the instrumental fingerprint. In this work, the instrument-agnostic fingerprints obtained from the analysis of volatile organic compounds present in hazelnuts using GC-MS were used to evaluate the differences among samples from the Italian Piedmont region and other hazelnut samples of industrial interest from different regions of Italy and Turkey. The PCA revealed that the differences contained in the instrument-agnostic fingerprint were due to country of origin, growing region, storage time and conditions of each sample. Three classification models (SIMCA, PLS-DA, and SVM) were developed to distinguish Piedmont samples from the remaining ones. After testing several data pre-processing methods, PLS-DA and SVM were the best performing classification algorithms.