Non-destructive beef adulteration detection using hyperspectral imaging and independent component analysis.

Area of Science:

• Food Science and Technology
• Analytical Chemistry
• Spectroscopy

Background:

• Traditional meat adulteration detection methods are destructive, slow, and unsuitable for rapid screening.
• Hyperspectral imaging (HSI) offers non-destructive detection but struggles with chemically similar adulterants.
• Artificial reconstruction in adulterated meat creates distinct surface morphology differences from authentic meat.

Purpose of the Study:

• To develop a novel non-destructive approach for detecting beef adulteration using HSI and ICA.
• To extract tissue surface features for differentiating authentic and adulterated beef.
• To investigate the detection of restructured minced meat and pork-substituted beef at various levels.

Main Methods:

• Combined hyperspectral imaging (HSI) with independent component analysis (ICA) to extract tissue surface features.
• Utilized derivatives to characterize surface-change trends for adulteration detection.
• Applied k-nearest neighbor, linear discriminant analysis (LDA), hierarchical cluster analysis (HCA), and orthogonal partial least squares discriminant analysis (OPLS-DA) for classification and quantification.

Main Results:

• LDA achieved perfect discrimination (R²p = 1.0000) for restructured minced beef detection.
• OPLS-DA demonstrated excellent performance (AUC > 0.92) for pork-substituted beef detection, outperforming HCA.
• Visualization successfully located pork-substituted regions and quantified substitution levels.

Conclusions:

• The developed tissue surface feature-based approach is rapid, accurate, and non-destructive for meat adulteration detection.
• This method shows significant potential for industrial implementation, especially for high-throughput screening.
• The technique is well-suited for quality assurance operations in meat processing facilities.