Feasibility of Using Fourier Transform Infrared Spectroscopy Combined with Chemometrics for Determing the Total Flavonoid Contents in Rape Bee Pollen

The potential of Fourier transform infrared spectroscopy (FTIRS) combined with chemometrics for its ability to fast determine the total flavonoids content of rape bee pollen was studied. A total of 25 rape bee pollen samples were investigated for FTIRS on the wavenumber between 7800 and 350 cm^-1. Direct orthogonal signal correction (DOSC) was performed to preprocess the original spectra data. Partial least squares (PLS) and kernel partial least square (KPLS) analysis were performed on the preprocessed spectral data using full spectrum data and 15 sensitive wavenumber spectra data extracted by DOSC. The prediction results show that the PLS and KPLS regression methods can be successfully employed to predict total flavonoids in rape bee pollen, the best prediction statistical parameters were 0.892 for correlation coefficient (R_P²) and 0.883 for root mean square error of prediction (RMSEP) by KPLS model. It is concluded that Fourier transform infrared spectroscopy combined with chemometrics has the potential to be used for fast determining the total flavonoids of rape bee pollen.