Prediction of protein and oil contents in soybeans using fluorescence excitation emission matrix

Abstract To investigate the potential of fluorescence spectroscopy in evaluating soybean protein and oil content, excitation emission matrix (EEM) was measured on 34 samples of soybean flours using a front-face measurement, and the accuracy of the protein and oil content prediction was evaluated. The EEM showed four main peaks at excitation/emission (Ex/Em) wavelengths of 230/335, 285/335, 365/475, and 435/495 nm. Furthermore, second derivative synchronous fluorescence (SDSF) spectra were extracted from the EEMs, and partial least square regression and support vector machine models were developed on each of the EEMs and SDSF spectra. The R2 values reached 0.86 and 0.74 for protein and oil, respectively. From the loading spectra, fluorescence at Ex/Em of 230–285/335 nm and 350/500 nm mainly contribute to the protein and oil content prediction, respectively. Those results revealed the potential of fluorescence spectroscopy as a tool for a rapid prediction of soybean protein and oil content.