Repositori URV

Identificador: imarina:9379684

Handle: https://hdl.handle.net/20.500.11797/imarina9379684

Autores:
Mendez-Sanchez, CGüell, MCFerrando, MJimenez-Flores, RCastellvi, SDRodriguez-Saona, LDomingo, JC

Resumen:
The present study describes a new approach to predict the level of crude fat present in commercial edible Tenebrio molitor and Alphitobius diaperinus powders partially defatted using mid and near infrared spectral data combined with multivariate analysis. Insect powders were partially defatted by using three organic solvents and CO2 in supercritical conditions obtaining samples with fat content ranging from 0 to 28.7% (n = 46). Lipid content and fatty acid profile were determined by using Soxhlet and fatty acid methyl esters (FAME) methods, respectively. Spectral data was acquired using a two handheld FT-NIR devices (1350-2550 nm) and two portable FT-MIR equipment (4000-630 cm- 1) equipped with ATR crystals. Partial least squares regression (PLSR) model was used to easily predict insect fat content. Ethanol had lipid extraction yields significantly lower, specially for T. molitor. FA composition was affected by the solvent used. PLSR results exhibited good linearity, predicting crude fat content with strong correlation (Rcv >= 0.9) and low standard error of cross-validation (SECV = 1.06-3.22%). Nonetheless, the FT-NIR devices tested, showed higher performance for fat content prediction in insect powders, reaching values of 0.99 in coefficient of correlation (RP) and 1.05% in standard error in prediction (SEP).