[6-1015-C] Postharvest/Food Technology and Process Engineering (6)

[6-1015-C-04] Application of Non-destructive Determination of Rice Amylose Content at Grain Elevators

*Edenio Olivares Diaz¹, Shuso Kawamura¹, Miki Matsuo¹, Toru Nagata², Shigenobu Koseki¹ (1. Hokkaido University(Japan), 2. Hokkaido Research Organization Central Agricultural Experiment Station(Japan))

キーワード：Oryza sativa L., Amylose content, Near-infrared spectroscopy, Chemometric analyses, Quality, Practical application

Rice (Oryza sativa L.) is the most important staple food for people in a large part of the world. Starch, moisture, and protein are the major constituents comprising the rice endosperm. However, amylose content (AC), which is the percentage of amylose relative to total starch in the rice kernel, contributes to the texture and quality of cooked rice. Iodine-binding, also known as iodine colorimetry or amylose-iodine, is the only validated and most commonly used method for determining AC. But it is labor-intensive, time-consuming, chemical-dependent, and vulnerable to random error. Therefore, it is unsuitable for laboratory and/or industrial uses where large volumes of samples need to be processed. To overcome this shortcoming, near-infrared (NIR) spectroscopy in combination with chemometric techniques represents an alternative, validated method for assessing rice AC. In this study, we developed an accurate model for the non-destructive determination of AC at grain elevators. A dual-step calibration model was developed using data from 936 samples of 10 varieties of rice produced between 2008 and 2018 in various regions of Hokkaido, Japan. The collected rough rice samples were dried to approximately 15% w.b. of moisture content. Next, each dried sample was hulled to obtain brown rice. Finally, each brown rice sample was milled to 90.5 ± 0.2% of milling degree. Milled rice AC reference values (AC_ref), transmittance NIR spectra, and physicochemical properties were combined to develop the dual-step calibration model. The raw NIR transmittance spectra was preprocessed by applying a 2^nd order Savitzky-Golay derivative with a 2^nd polynomial order. Later, AC_ref values, transformed NIR spectra, and physicochemical properties were analyzed by partial least squares regression (PLS) and multiple linear regression (MLR) to develop an accurate dual-step calibration model. Our dual-step calibration model described low and ordinary amylose varieties models, which were developed individually. Both the low and ordinary amylose varieties models were calibrated using 2008-2017 production year samples (n = 903) and validated by distinct 2018 production year samples (n = 33), which were collected at a rice grain elevator. Next, the dual-step calibration model was created by merging the validation results of the low and ordinary amylose varieties models. Results indicated that AC_ref values were determined with high accuracy based on the low average standard error of the laboratory method (SEL) = 0.17% among varieties and production years. Also, the regression coefficients of each wavelength related to AC_ref for the optimal PLS factor indicated that the wavelength at 916 nm reported the highest spectral variation and thus correlated the most to AC. Moreover, validation statistics such as standard error of prediction (SEP) = 0.33% and ratio of performance deviation (RPD) = 5.09 indicated the high robustness and accuracy of the dual-step model, enabling more precise, accurate, and efficient rice quality screening at Japanese grain elevators.