Skip to main content
ARS Home » Research » Publications at this Location » Publication #146878
Title: A COMPARATIVE STUDY OF FOURIER TRANSFORM RAMAN AND NIR SPECTROSCOPIC METHODS FOR ASSESSMENT OF PROTEIN AND AMYLOSE IN RICE

Author
item SOHN, MI RYEONG - USDA-FAS
item HIMMELSBACH, DAVID
item BARTON II, FRANKLIN

Submitted to: Cereal Chemistry
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/13/2004
Publication Date: 7/1/2004
Citation: Sohn, M., Himmelsbach, D.S., Barton II, F.E. 2004. A Comparative Study of Fourier Transform Raman and NIR Spectroscopic Methods for Assessment of Protein and Amylose in Rice. Journal of Cereal Chemistry. 81(4): 429-433.

Interpretive Summary: In a previous study we developed robust calibration models for determining rice cooking quality parameters such as apparent amylose and protein using near-infrared (NIR) spectroscopy, here two-years data were used and good results were obtained by the use of derivative method. This study designed to build robust calibration models using FT-Raman spectroscopy and to compare the precision between two spectroscopic methods, FT-Raman and NIR. The same sample set and same standardization method were used for exact comparison, and the effect of various chemometric analyses for modeling was investigated. The results indicated that both FT-Raman and NIR spectroscopic technique are potential methods to develop calibration model for rice quality using multi-year data and presented the best chemometric method.

Technical Abstract: Near infrared (NIR) reflectance spectroscopy and Fourier-transform Raman (FT-Raman) spectroscopy were used to compare the robust calibration models for determining rice quality parameters such as amylose and protein. The sample set used in calibration contains two-years of data with a wide range of constituent values. The Two-years spectral data for both FT-Raman and NIR were preprocessed with orthogonal signal correction (OSC) for standardization. For both spectroscopic methods, five models were optimized by partial least squares regression (PLSR) and Martens' uncertainty regression (MUR), including no processing, smoothing, normalization, and first- (D1) and second (D2) derivative. The NIR method was slightly better than the FT-Raman method in model performance for amylose, whereas the FT-Raman method was superior to the NIR for protein. MUR nearly did not offer improvement results over the PLSR for both constituents and for both methods. The best FT-Raman models were generated from OSC preprocessing for protein (SECV=0.15%, 5 factors) and for amylose (SECV=0.67%, 7 factors). The best NIR models were obtained with D2 transform of OSC spectra for protein (SECV=0.22%, 4 factors) and with OSC for amylose (SECV=0.57%, 11 factors).