Combined analysis of near-infrared spectra, colour, and physicochemical information of brown rice to develop accurate calibration models for determining amylose content. (15th July 2019)

• Record Type: Journal Article
• Authors: Olivares Díaz, Edenio; Kawamura, Shuso; Matsuo, Miki; Kato, Mizuki; Koseki, Shigenobu
• Is Part Of: Food chemistry. Volume 286(2019)
• Page Start: 297
• Page End: 306
• Publication Date: 2019-07-15
• Subjects: Rice quality -- Amylose content -- Near-infrared spectroscopy -- Calibration model accuracy -- Chemometric techniques; Food -- Analysis -- Periodicals; Food -- Composition -- Periodicals; 664
• DOI: 10.1016/j.foodchem.2019.02.005
• Languages: English
• ISSNs: 0308-8146

Abstract

Graphical abstract: Highlights: Development of models to assess amylose non-destructively using brown rice data and chemometrics. Models developed by merging the low and ordinary amylose levels validation results. Accuracy of developed models was suitable for industrial application. Combination of NIR spectra and physicochemical data revealed the most robust model. Contribution to more precise rice quality screening at grain elevators. Abstract: Amylose content is an important determinant of rice quality. Accurate non-destructive determination of amylose content remains a primary challenge for the rice industry. Here, we analysed the accuracy of three models for the non-destructive determination of amylose content. The models were developed by combining near-infrared spectra, colour, and physicochemical information relative to 832 brown rice samples from ten varieties produced between 2009 and 2017 in various regions of Hokkaido, Japan. Models describing low and ordinary amylose varieties were developed individually, merged, and validated using production year samples (2016–2017) different from the calibration set (2009–2015). The resulting accuracy was suitable for industrial application. With standard error of prediction = 0.70% and ratio of performance deviation = 3.56, the combination of near-infrared spectra and physicochemical information produced the most robust model, enabling more precise rice quality screening at grain elevators.