Near infrared reflectance spectroscopy of pasticceria foodstuff as protein content predicting method

Background: The authors evaluated the potential of NIR spectroscopy for the analysis of protein of pasticceria food, and it have analyzed 120 samples. Methods: Protein content reference was obtained by standard lab procedure. The authors used spectra and reference data, partial least squares regression analysis that was applied to calculate a NIR method model (n=120), divided into calibration (n=84) and validation (n=36), to predict protein content in fed. Results: The study shows that using NIR technique spectra and using PLS regression, a model could be developed having a root mean square standard error of performance of R2 value of 0.81. The regression coefficients show that analytically useful absorptions for the original PLS model are between 4000 and 4500nm. Conclusion: a NIR model was developed for the prediction of protein in pasticceria products, which is sufficiently accurate for screening samples. Screening can be accomplished rapidly without the use of chemicals.