Spectral markers and machine learning: Revolutionizing Rice evaluation with near infrared spectroscopy.

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Food Science

Background:

• Traditional rice variety evaluation is complex and time-consuming.
• Advanced equipment is often required for accurate rice quality assessment.
• There is a need for high-throughput methods to analyze rice properties.

Purpose of the Study:

• To discriminate 22 commercial rice varieties from six types.
• To analyze biochemical, physicochemical, and cooking properties of rice.
• To develop a high-throughput method for rice variety evaluation.

Main Methods:

• Near-infrared (NIR) spectroscopy was employed.
• Machine learning algorithms, including Partial Least Squares (PLS) and Principal Component Analysis (PCA), were utilized.
• Partial Least Squares Discriminant Analysis (PLS-DA) was used for classification.

Main Results:

• PLS models accurately predicted quality traits like whiteness (R² = 0.94), width (R² = 0.94), resilience (R² = 0.96), and springiness (R² = 0.98).
• PCA revealed distinct clustering patterns among rice varieties.
• PLS-DA achieved a 17% error rate in external predictions, identifying key spectral markers.

Conclusions:

• NIR spectroscopy combined with machine learning offers a high-throughput solution for rice variety discrimination.
• This approach enables precise quantification, classification, and differentiation of rice types.
• The method can enhance quality control, consumer satisfaction, and breeding selection processes.