Integration of surface-enhanced Raman spectroscopy (SERS) and machine learning tools for coffee beverage...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Chemometrics

Background:

• Surface-enhanced Raman spectroscopy (SERS) offers powerful molecule identification capabilities.
• Complex samples present challenges in SERS due to overlapping peaks and signal variability from non-uniform substrates.
• Machine learning (ML) techniques can address SERS data complexity, similar to facial recognition applications.

Purpose of the Study:

• To develop and evaluate a sensor system for classifying coffee beverages.
• To integrate SERS, feature extraction, and ML classifiers for robust analysis.
• To demonstrate a practical quality-control tool for the food industry.

Main Methods:

• Utilized a low-cost nanopaper substrate for enhanced SERS signals of coffee compounds.
• Employed Principal Component Analysis (PCA) and Discriminant Analysis of Principal Components (DAPC) for feature extraction.
• Evaluated various ML classifiers, including Support Vector Machine (SVM) and K-Nearest Neighbor (KNN).

Main Results:

• The combination of DAPC with SVM or KNN achieved the highest performance in coffee beverage classification.
• Nanopaper demonstrated effectiveness as a versatile SERS substrate for dilute analytes.
• The integrated system successfully addressed challenges of peak overlap and signal variability.

Conclusions:

• The developed sensor system provides a user-friendly and versatile approach for coffee classification.
• The integration of SERS with DAPC and ML classifiers offers a promising solution for complex sample analysis.
• This technology has significant potential as a practical quality-control tool in the food industry.