Streamlining Group Type Analysis with Standard GCxGC Templates through Computer Vision-Assisted Alignment

Significance of the Topic

Comprehensive two-dimensional gas chromatography (GC×GC) has become a cornerstone technique for detailed hydrocarbon group type analysis in petroleum products. By overcoming the limitations of one-dimensional GC, GC×GC enables clearer separation of overlapping compounds, supports regulatory compliance, and enhances quality control in industrial and environmental applications.

Goals and Study Overview

This work introduces a template-based workflow that leverages computer vision to automate retention time (RT) alignment for group type analysis. Key objectives include:

• Establishing a robust method to match reference peaks via local chromatographic patterns.
• Developing global and local transformations to align standard GC×GC templates to new chromatograms.
• Validating alignment accuracy and group boundary verification through pattern correlation.

Methodology and Used Instrumentation

This approach integrates the following steps:

• Visual peak matching using Enhanced Correlation Coefficient (ECC) to compare local 2D chromatographic patterns.
• Calculation of affine and polynomial transformations (global) complemented by region-specific functions (local) to map RT shifts.
• Verification of group boundaries by computing normalized correlation scores between template and sample patterns.

Instrumentation referenced includes flow-modulated GC×GC with flame ionization detection (FID) as specified in ASTM D8396 and UOP 990 methods. Custom software from GC Image, LLC performs pattern extraction, ECC computation, transformation fitting, and scoring.

Main Results and Discussion

Evaluation on 166 chromatograms with over 7 600 peaks showed that the correct reference peak ranked first in 82.2% of cases. Expanding to top-5 and top-10 candidates increased correct matches to 92.5% and 94.7%, respectively. Inter-lab and intra-lab examples demonstrated that combined global and local transformations effectively realign templates across different systems and over time. Low correlation scores flagged mismatches, reducing manual review.

Benefits and Practical Applications

This workflow offers:

• Automated RT alignment to accelerate routine hydrocarbon typing.
• Reduced manual intervention through ranked match candidates and confidence scoring.
• Enhanced reproducibility across laboratories and over extended timeframes.

Future Trends and Potential Applications

Advances may include integration of deep-learning pattern recognition, real-time adaptive alignment during acquisitions, and extension to other complex mixtures (e.g., biofuels, environmental samples). Cloud-based template libraries and collaborative alignment databases could further streamline method sharing.

Conclusion

By combining computer vision–based peak matching with flexible RT transformations, this workflow enables reliable, automated alignment of GC×GC templates for hydrocarbon group type analysis. The method achieves high matching accuracy, supports quality assurance through correlation scoring, and promises to reduce manual effort in routine analyses.

References

• Psarakis E.Z., Maragos P. (2005) Enhanced Correlation Coefficient Image Alignment.
• Rempe D. et al. (2016) Anal. Chem., application of GC×GC over multi-year and multi-run datasets.
• ASTM D8396-22, Standard Test Method for Group Types Quantification by GC×GC–FID.
• UOP 990-11, Organic Analysis of Distillate by Comprehensive Two-Dimensional GC with FID.