Wet Gluten Rapid Detection Assay

A non-destructive rapid assay for wet gluten content in wheat, based on visible/NIR hyperspectral data and random forest regression, with accuracy and robustness surpassing conventional methods.

Li, Yan · Sha, Min · Li, Peng · Zhang, Zhengyong

Foods 2025

Specifications

Coefficient of determination

{'zh': '0.8579', 'en': '0.8579'}

Root mean square error

{'zh': '0.0216', 'en': '0.0216'}

Relative percent deviation

{'zh': '2.6978', 'en': '2.6978'}

Coefficient of determination

{'zh': '0.8383', 'en': '0.8383'}

Root mean square error

{'zh': '0.0231', 'en': '0.0231'}

Relative percent deviation

{'zh': '2.5293', 'en': '2.5293'}

Coefficient of determination

{'zh': '0.8474', 'en': '0.8474'}

Root mean square error

{'zh': '0.0224', 'en': '0.0224'}

Advantages

No chemical reagents

Directly acquires spectra to predict wet gluten content, eliminating solvent preparation, sample pretreatment, and waste disposal, suitable for workshops and procurement sites.

Handles both grains and flour

The same pipeline is validated on both grains and flour, so grain procurement and flour processing can share the model without separate development.

More accurate content prediction

After first-derivative and SG smoothing, visible spectra fed to random forest regression achieve an r² of 0.8579 with RMSE of 0.0216.

Applications

• Rapid wheat quality testing：Procurement sites scan spectra to obtain wet gluten values in seconds, enabling quality-based pricing without lab delays.
• Flour wet gluten determination：Flour mills predict wet gluten online to adjust wheat blending and tempering parameters in real time.
• Non-destructive food component detection：Non-destructive measurement of gluten without sample preparation or reagents, suitable for rapid component detection in food.
• Hyperspectral analyzer：Embeds the wet gluten prediction model into hyperspectral analyzers, enabling them to output content results directly.