Miniaturized NIR and MIR spectroscopy for insect lipid characterization

Abstract The main objective of this research was to evaluate the performance of portable FT-MIR (4000 to 630 cm −1 ) and handheld FT-NIR (7400 to 4000 cm −1 ) spectrometers, combined with partial least squares regression (PLSR) multivariate analysis. This method was used to predict the percentage of insect lipids added to mixtures of vegetable and seed oils, as well as to profile the fatty acid composition in the lipid fractions of Locusta migratoria and Acheta domesticus mixed with these oils. Insect lipids from L. migratoria and A. domesticus were extracted from their powdered forms using organic solvent extraction and then mixed with five edible oils (olive, sunflower, avocado, almond, and linseed) at concentration ratios ranging from 0 to 100% (n = 140). The fatty acid (FA) profile of the samples was determined using the fatty acid methyl ester (FAME) method. PLSR was used to develop algorithms for predicting insect lipid content in mixtures and FA composition. The PLSR model for predicting the percentage of insect lipids in mixtures with vegetable and seed oils demonstrated good linearity, with high correlation ( R CV = 0.99), low standard errors of cross-validation (SE CV ≤ 2.21%) and prediction (SE P ≤ 2.32%), and high predictive performance as indicated by the ratio of prediction to deviation (RPD ≥ 9.23). The percentage of myristic, palmitic, stearic, oleic, linoleic, and α-linolenic fatty acid as well as the content of saturated (SFA), monounsaturated (MUFA), and polyunsaturated (PUFA) were accurately predicted with values of SE CV ≤ 4.64% and an R CV ≥ 0.88. However, the FT-MIR spectrometer tested showed superior performance in predicting fatty acids and SFA, MUFA, PUFA reaching values of 0.99 in coefficient of correlation (RP) and 0.66% in standard error in prediction (SE P ).