This study aimed to test the feasibility of applying FTIR spectroscopy in order to create models that can predict quality attributes of pea composites. FTIR spectroscopy data were captured from three types of samples namely pea flour (PF), pea concentrate (PC) and pea isolate (PI). The FTIR spectral data along with partial least square (PLS) regression were used to build predictive models for physicochemical (moisture content, protein content, starch content, fiber content, bulk density, color), structural (particle size distribution, surface openings, fractal dimension), thermal (denaturization temperature, enthalpy), and techno-functional traits (water holding capacity, foaming capacity, foam stability, solubility, least gelation concentration) of pea composites. The FTIR spectra were subjected to different spectral pretreatment where 2nd derivative pretreatment provided the most suitable model for prediction of the studied parameters. The values of pea composite's traits determined through non-destructive FTIR spectroscopy coupled with partial least squares regression analysis, were very closed to the values obtained by the destructive standard methods. Performance (correlation, root mean square error) of the developed models were attribute-specific. For most of the studied attributes, correlation coefficient (r) were higher than 0.82 in the calibration step, and 0.71 in the prediction step. Pea composites have shown distinctive functionalities both as individual entity and in formulating meat analogs. Overall, it could be recommended that FTIR spectroscopy could be used in pea processing industry for developing robust models, to non-destructively assess the pea composite's properties and techno-functionalities.
Read full abstract