Food Classification and Authenticity Testing Using a New High-Resolution LC/QTOF and Novel Classification Software

Importance of the Topic

Food adulteration is a critical challenge in quality control and safety of food products. High-resolution mass spectrometry combined with multivariate analysis provides a powerful approach to verify authenticity and detect contaminants in complex food matrices.

Aims and Study Overview

This study demonstrates a streamlined workflow to classify mango varieties and detect adulteration using high-resolution LC/Q-TOF instrumentation and the new MassHunter Classifier 1.0 software. A proof-of-concept model was developed with three mango types—Keitt, Ataulfo, and Tommy Atkins—and evaluated for routine use by laboratory technicians.

Methodology and Instrumentation

Sample Preparation and Extraction:

• Six fruits per variety were homogenized and extracted with QuEChERS EN kits.
• A pooled positive quality control per variety was prepared and mixed at specific ratios to generate adulterated samples.
• A deuterated pesticide (dimethoate-D6) at 50 ppb was added as an internal standard to monitor injection consistency.

Chromatography and Detection:

• Agilent 1290 Infinity II LC coupled with Agilent 6546 LC/Q-TOF.
• Acquisition of full spectra from m/z 50–1000 without daily recalibration.

Data Processing and Modeling:

• Feature extraction using MassHunter Profinder 10.0 with batch recursive algorithms.
• Data filtering and statistical analysis in Mass Profiler Professional 15.0.
• PLS-DA model construction and export to MassHunter Classifier 1.0 for simplified routine analysis.

Main Results and Discussion

The 6546 LC/Q-TOF platform delivered stable performance with <6% internal standard area variation and <2 ppm mass error across all injections. Profinder identified 4,185 initial features across the three mango varieties, which were narrowed to 481 robust markers through stringent statistical thresholds. The PLS-DA model achieved perfect classification of pure and adulterated samples using a 0.8 confidence threshold, with 100% accuracy on both day 1 and day 14 evaluation sets. MassHunter Classifier enabled technicians to import the model file and sample data via a simple interface, producing rapid categorization supported by confidence scores and PCA plots indicating group membership and outliers.

Benefits and Practical Applications

The proposed workflow reduces reliance on specialist chemometric software for routine testing, enabling lab personnel to perform authenticity checks with minimal training. The high-resolution LC/Q-TOF system ensures data consistency over longitudinal studies, while the MassHunter Classifier interface accelerates decision-making through intuitive visualization of confidence scores and PCA clustering.

Future Trends and Opportunities

Expanding model libraries to encompass a broader range of food products and geographical origins. Incorporating advanced machine learning algorithms to improve model robustness and adapt to sample variability over time. Automating end-to-end workflows for real-time monitoring of supply chains and rapid detection of novel adulterants.

Conclusion

This study establishes an efficient, high-throughput pipeline for food authenticity testing using high-resolution LC/Q-TOF and dedicated classification software. The combined approach offers accuracy, reproducibility, and ease of use suitable for routine quality control in food laboratories.