Optimization of Near‐Infrared Reflectance Model in Measuring Gelatinization Characteristics of Rice Flour with a Rapid Viscosity Analyzer (RVA) and Differential Scanning Calorimeter (DSC)

This study compared the calibration models generated by combinations of different mathematical and preprocessing treatments as well as regression algorithms to optimize the analysis of gelatinization properties of rice flour by using near‐infrared spectroscopy, in comparison with conventional techniques of differential scanning calorimetry (DSC) and rapid viscosity analysis (RVA). A total of 220 milled rice flours were used for model construction. A model generated by the modified partial least squares regression (MPLS) with mathematical treatment “2, 8, 8, 2” (second‐order derivative computed based on eight data points, and eight and two data points in the second smoothing, respectively) and detrend preprocessing was identified as the best for simultaneously measuring onset temperature (T_o), peak temperature (T_p), and conclusion temperature (T_c) of DSC. MPLS/“2, 8, 8, 2”/weighted multiplicative scattering correction preprocessing was identified as the best for RVA properties. The results indicated that near‐infrared reflectance spectroscopy could be used to rapidly predict gelatinization properties of rice flour for the purposes of quality evaluation of germplasm and selection of intermediate lines in breeding programs.