Use of MALDI-TOF mass spectrometry and machine learning to detect the adulteration of extra virgin olive oils

Significance of the Topic

Extra virgin olive oil (EVOO) is valued for its nutritional and economic importance, making it prone to adulteration with lower-grade seed oils. Rapid, reliable analytical tools are essential to safeguard product quality and consumer trust.

Objectives and Study Overview

This study aimed to develop a straightforward workflow combining MALDI-TOF mass spectrometry profiling of triacylglycerols (TAGs) with a neural network-based logistic regression model to detect sunflower oil adulteration in EVOO at levels down to 5%.

Methodology and Instrumentation

• Sample Preparation: EVOO and sunflower oil solutions in chloroform spiked with tricaprin internal standard; mixtures prepared at 0–20% sunflower oil.
• MS Analysis: Matrix-free LDI-TOF MS on a Shimadzu MALDI-8020 benchtop spectrometer; quadruplicate acquisitions for each sample.
• Data Processing: Peak alignment and area calculations via Clover MS software; normalization against m/z 903 Da.
• Machine Learning: Three-layer neural network with logistic regression, trained on 267 spectra, validated across 45 spectra, and tested on 30 blinded samples.

Results and Discussion

Distinct TAG profiles were observed for EVOO and sunflower oil, notably in ratios such as m/z 877/907, 901/907, 903/907, and 905/907. These ratios exhibited strong linear correlations (R2 > 0.996) with sunflower oil content. The neural network classifier achieved 97.8% validation accuracy and correctly classified all 30 blinded samples (100%).

Benefits and Practical Applications

• Fast and minimal sample preparation without chemical matrix.
• High sensitivity to detect low-level adulteration.
• Potential integration into routine quality control workflows.

Future Trends and Potential Applications

Expanding the spectral database to include diverse oil types and refining machine learning models could enhance robustness. Portable MALDI-TOF instruments and real-time classification may enable on-site authentication in production facilities and supply chains.

Conclusion

The combined MALDI-TOF MS and neural network approach offers a rapid, accurate method for detecting EVOO adulteration. Further validation and dataset expansion will support industrial implementation.

References

No external literature references were cited.