Determination of protein, lipid and sugar contents of bakery products by using Fourier-transform near infrared spectroscopy

The use of Fourier-transform near infrared spectroscopy (FT-NIR) to measure the content of protein, lipid and sugar contents of bakery products was investigated. The samples were dried, homogenized, sieved and measured in the wavelength range of 780–2500 nm. The calibration was based on partial least squares (PLS) regression with cross-validation. The performance of the final model was evaluated according to root mean square of cross-validation (RMSECV), root mean square error of estimation (RMSEE) and the determination coefficient (R2).The developed models use the ranges of 1100–1245 nm and 1590–2600 nm for protein determination, 1330–1840 nm and 2170–2350 nm for lipid, 1400–1630 nm, 2000–2170 nm and 2230–2570 nm for sugar determination, respectively. Protein, lipid and sugar could be determined directly with R2 values of 98.93, 99.07 and 98.81, and RMSECV values of 0.16 m/m%, 0.79 m/m% and 0.28 m/m%, respectively. It can be concluded that FT-NIR spectroscopy can be used for the routine determination of protein, lipid and sugar content of bakery products and it can contribute to the estimation of calorie content in a fast and non-destructive way.