Modeling Textural Properties of Cooked Germinated Brown Rice Using the near-Infrared Spectra of Whole Grain

Area of Science:

• Food Science and Technology
• Agricultural Engineering
• Analytical Chemistry

Background:

• Developing non-destructive, rapid techniques for assessing cooked germinated brown rice (GBR) texture is crucial for commercial quality control.
• Near-infrared (NIR) spectroscopy offers a promising avenue for such analyses due to its speed and non-invasive nature.

Purpose of the Study:

• To develop and validate a non-destructive method for evaluating the textural properties (hardness, toughness, adhesiveness) of cooked germinated brown rice (GBR).
• To compare the performance of artificial neural network (ANN) and partial least squares (PLS) regression models using NIR spectral data.

Main Methods:

• Fourier transform near-infrared (NIR) spectral data from uncooked whole grain GBR were collected.
• Artificial neural network (ANN) and partial least squares (PLS) regression were employed for data analysis.
• Data separation (Kennard-Stone method) and spectral pretreatment (smoothing, standard normal variate) were investigated to optimize models.

Main Results:

• The ANN model demonstrated high accuracy in predicting hardness (R²=1.00, r²=0.94), toughness (R²=1.00, r²=0.92), and adhesiveness (R²=0.97, r²=0.91) of cooked GBR.
• PLS regression models also provided good predictions for hardness and toughness, with r² values of 0.85 and 0.82, respectively.
• Specific NIR spectral regions and vibration bands related to amylose were identified as influential for texture prediction.

Conclusions:

• The developed ANN model using NIR spectroscopy is a viable and effective tool for the rapid, non-destructive assessment of cooked GBR textural properties.
• This technique holds significant potential for implementation in GBR production factories for quality assurance and product formulation.
• Further refinement with more diverse samples and brands can enhance the robustness of the predictive models.