Use of color based chromatographic images obtained from comprehensive two-dimensional gas chromatography in authentication analyses

Comprehensive two-dimensional gas chromatography (GCxGC) has been an important technique used to acquire as much information as possible from a wide variety of samples. Qualitative contour plots analysis provides useful information and in daily use it ends up being handled as images of the volatile organic compounds by analysts. Cachaca samples are used in this paper to showcase the use of two-dimensional chromatographic images as the main source for authentication purposes through one-class classifiers, such as data-driven soft independent modeling of class analogy (DD-SIMCA). The proposed workflow summarizes this fast and easy process, which can be used to certify a specific brand in comparison to other brands, as well as to authenticate if samples have been adulterated. Lower quality cachacas, non-aged cachacas and cachacas aged in different wooden barrels were tested as adulterants. Chromatographic images allowed for the distinction of all brands and nearly every adulteration tested. Sensitivity was estimated at 100% for all models and specificity ranged from 96% to 100%. Different approaches were used, alternating from working with whole-sized images to working with smaller resized versions of those images. Resized chromatographic images could be potentially useful to easily compensate for slight chromatographic misalignments, allowing for faster calculations and the use of simpler software. Reductions to 50% and 25% of the original size were tested and the results did not greatly differ from whole images model. As such, 2D chromatographic images have been found to be an interesting form of evaluating a product's authenticity.

Automated summary

The study demonstrates the use of 2D chromatographic images as a reliable method for authenticating samples, showing high sensitivity and specificity in distinguishing between different brands of cachacas and adulterants. By utilizing qualitative contour plots analysis and one-class classifiers, the proposed workflow effectively summarizes a fast and easy process for certification and authentication purposes. The use of resized chromatographic images is also explored, showing potential for faster calculations and simpler software usage without significant loss of accuracy.