Protein and apparent amylose contents of milled rice by NIR-FT/Raman spectroscopy

The chemometric calibration of near-infrared Fourier-transform Raman (NIR-FT/Raman) spectroscopy was investigated for the purpose of providing a rigorous spectroscopic technique to analyze rice flour for protein and apparent amylose content. Ninety rice samples from a 1996 collection of short, medium, and long grain rice grown in four states of the United States, as well as Taiwan, Korea, and Australia were investigated. Milled rice flour samples were scanned in rotating cups with a 1,064 nm (NIR) excitation laser using 500 mW of power. Raman scatter was collected using a liquid N-2 cooled Ge detector over the Raman shift range of 175-3,600 cm(-1). The spectral data was preprocessed using baseline correction with and without derivatives or with derivatives alone and normalization. Nearly equivalent results were obtained using all of the preprocessing methods with partial least squares (PLS) models. However, models using baseline correction and normalization of the entire spectrum, without derivatives, showed slightly better performance based on the criteria of highest r(2) and the lowest SEP with low bias. Calibration samples (n = 57) and validation samples (n = 33) were chosen to have similar respective distributions for protein and apparent amylose. The best model for protein was obtained using six factors giving r(2) = 0.992, SEP = 0.138%, and bias = -0.009%. The best model for apparent amylose was obtained using eight factors giving r(2) = 0.985, SEP = 1.05%, and bias = -0.006%.