Lipids in a Nutshell: Quick Determination of Lipid Content in Hazelnuts with NIR Spectroscopy.
Elena CazzanigaNicola CavalliniAlessandro GiraudoGentian GavociFrancesco GeobaldoMattia ParianiDaniela GhirardelloGiuseppe ZeppaFrancesco SavoraniPublished in: Foods (Basel, Switzerland) (2022)
Hazelnuts ( Corylus avellana L.) are among the most consumed dry fruits all over the world. Their commercial quality is defined, above all, by origin and dimension, as well as by lipid content. Evaluation of this parameter is currently performed with chemical methods, which are expensive, time consuming, and complex. In the present work, the near-infrared (NIR) spectroscopy, using both a benchtop research spectrometer and a retail handheld instrument, was evaluated in comparison with the traditional chemical approach. The lipid content of hazelnuts from different growing regions of origin (Italy, Chile, Turkey, Georgia, and Azerbaijan) was determined with two NIR instruments: a benchtop FT-NIR spectrometer (Multi Purpose Analyser-MPA, by Bruker), equipped with an integrating sphere and an optic fibre probe, and the pocket-sized, battery-powered SCiO molecular sensor (by Consumer Physics). The Randall/Soxtec method was used as the reference measurement of total lipid content. The collected NIR spectra were inspected through multivariate data analysis. First, a Principal Component Analysis (PCA) model was built to explore the information contained in the spectral datasets. Then, a Partial Least Square (PLS) regression model was developed to predict the percentage of lipid content. PCA showed samples distributions that could be linked to their total crude fat content determined with the Randall/Soxtec method, confirming that a trend related to the lipid content could be detected in the spectral data, based on their chemical profiles. PLS models performed better with the MPA instrument than SCiO, with the highest R 2 of prediction (R 2 PRED = 0.897) achieved by MPA probe, while this parameter for SCiO was much lower (R 2 PRED = 0.550). Further analyses are necessary to evaluate if more acquisitions may lead to better performances when using the SCiO portable spectrometer.
Keyphrases
- fatty acid
- data analysis
- high resolution
- photodynamic therapy
- fluorescence imaging
- fluorescent probe
- drug release
- optical coherence tomography
- living cells
- adipose tissue
- machine learning
- patient reported outcomes
- magnetic resonance
- artificial intelligence
- big data
- density functional theory
- low cost
- molecular dynamics