Solvent-based mid-infrared spectroscopy paired with modern machine learning approaches for saffron authentication.

Area of Science:

• Analytical Chemistry
• Chemometrics
• Spectroscopy

Background:

• Saffron quality and authenticity are critical for its market value.
• Traditional methods for saffron analysis can be time-consuming and labor-intensive.
• Need for rapid, reliable, and non-destructive analytical techniques.

Purpose of the Study:

• To develop and validate a solvent-based mid-infrared (MIR) spectroscopy method for saffron quality control.
• To discriminate geographical origin of saffron samples.
• To detect adulteration of saffron with common plant-based adulterants.

Main Methods:

• Analysis of 111 authentic saffron samples from six Iranian regions using a MIRA analyzer.
• Solvent extraction protocol optimized using Design of Experiments (DOE).
• Application of chemometric techniques: Principal Component Analysis (PCA), Partial Least Squares-Discriminant Analysis (PLS-DA), Random Subspace Discriminant Ensemble (RSDE), and Data-driven Soft Independent Modeling of Class Analogy (DD-SIMCA).

Main Results:

• Geographical origin discrimination achieved 93.0% accuracy with PLS-DA and 95.5% with RSDE.
• 100% sensitivity and specificity in differentiating authentic from adulterated saffron using DD-SIMCA.
• RSDE identified adulterant types and levels with >94.0% accuracy, outperforming PLS-DA (>90.0%).

Conclusions:

• Solvent-based MIR spectroscopy combined with chemometrics is a powerful tool for saffron quality and authenticity control.
• The developed method offers high accuracy for geographical discrimination and adulteration detection.
• This approach shows potential for point-of-need saffron analysis.