Version française
RSS feeds
Accueil
Home > News > News stream... > A journey to the heart of dunes for products in quest of identity, or how image processing accelerates their characterization

News stream...

A journey to the heart of dunes for products in quest of identity, or how image processing accelerates their characterization

March 2018

IFP Energies nouvelles has developed BARCHAN, an analysis tool for robust registration between two 2D chromatographic images. Based on recent image processing and optimization techniques, BARCHAN significantly reduces the time - by a factor of more than 10 - required to process and quantify 2D chromatograms.

In the food, cosmetics or materials industries and in chemical, physical, biological and forensic sciences*, product analysis consists in identifying, characterizing and quantifying their various constituent elements. So-called “complex” mixtures refer to products made up of hundreds or thousands of different molecules, present in varying proportions.
  
Their detailed analysis requires the use of analytical tools and data processing algorithms. The purpose of the former is to separate the different constituent molecules as effectively as possible while the latter are used to process the data obtained, and from them, calculate the information needed to produce an identity card for each of the products.
    
Gas chromatography is one among many available analytical tools (such as spectrometry and spectroscopy, magnetic resonance, etc. [Lynch-2003]). Several variations thereof exist, including those combining different types of chromatography in a multidimensional fashion. Two-dimensional chromatography [Vendeuvre-2004] is a notable example: it is used to separate constituent molecules of hydrocarbon samples according to two complementary physicochemical properties,. The identity card it produces (a two-dimensional (2D) chromatogram) is shown in figure 1.

It is depicted in the form of a landscape made up of dunes and peaks, each corresponding to a particular molecule, whose properties and proportion are identified by its geographic coordinates and altitude. Peak alignments are characteristic of molecules from the same chemical family. In the first 2D chromatograms obtained at IFPEN in the early 2000s, some of the peaks had a curved appearance, forming landscapes reminiscent of sand deserts, with waves of crescent-shaped dunes, sculpted by the wind, known as barchans (they can even be found on the planet Mars).

While these geographic identity cards reveal precious information about the products analyzed, the proliferation of peaks impedes the comparison of two similar compounds from different samples. From one analysis to another, peaks may have shifted slightly and some can be masked by genuine mountains (corresponding to a molecule that is present in abundance in one product but not the other). Sometimes, a focus is laid on a string of very small aligned peaks, representing families of trace molecules. Comparing two chromatograms is then like comparing two partial maps of the same location, produced at two distinct geological points in time, a tedious manual process to identify comparable locations and recalculate their respective distances.

BARCHAN (Blob Alignment for Robust CHromatographic ANalysis [Couprie-2017]) is a tool for robust registration between two 2D chromatograms, based on recent image processing and optimization techniques. It is based on the automatic identification of the highest peaks (or blobs) on each of the chromatograms, and calculates the optimal continuous transformation (or morphing) enabling the superimposition of the two maps. This remains possible even if small peaks have become mountains, if the ground level (or baseline) has changed [Ning-2014], if the maps are not on the same topographical scale or if one of them presents entire regions that are unexplored (no peaks identified). This automated analysis is conducted in a matter of minutes compared to a manual process that previously took hours.    
Maps transformed in this way can then be retouched by hand by analytical chemists to precisely adjust some of the more subtle geographic details (a few specific molecules) that cannot be automatically calibrated. Ultimately, BARCHAN significantly reduces the time required - more than ten-fold - to process and quantify 2D chromatograms. This technique is also useful for comparing chromatograms from different chromatographs, thereby facilitating cross-referencing or comparison of results between research laboratories and teams.

    

This work is the most recent result from a decade-long collaboration, since IFPEN’s first PhD thesis on such methods [Vendeuvre-2004], between teams from the Physics and Analysis and Mechatronics, Computer Science and Applied Mathematics divisions.
   

* Forensic sciences refer to the analytical methods employed (imaging, chemistry, biology, computer science) in investigations to identify material clues and evidence. The word comes from the Latin word forum, where suspects accused of a crime were given a public hearing

 

Scientific contact:  laurent.duval@ifpen.fr  

Publications :

  1. [Couprie-2017] BARCHAN: Blob Alignment for Robust CHromatographic ANalysis (GCxGC), Camille Couprie, Laurent Duval, Maxime Moreaud, Sophie Hénon, Mélinda Tebib, Vincent Souchon, Journal of Chromatography A, 2017
    >> https://dx.doi.org/10.1016/j.chroma.2017.01.003
      
  2. [Ning-2014] Chromatogram baseline estimation and denoising using sparsity (BEADS), Xiaoran Ning, Ivan Selesnick, Laurent Duval, Chemometrics and Intelligent Laboratory Systems, December 2014
    >> https://dx.doi.org/10.1016/j.chemolab.2014.09.014
        
  3. BEADS: Baseline Estimation And Denoising with Sparsity - Known BEADS uses in data processing and signal trend removal/detrending - References and code: Matlab, R, C++, Laurent Duval
    >> http://www.laurent-duval.eu/siva-beads-baseline-background-removal-filtering-sparsity.html
        
  4. [Vendeuvre-2004] Comparison of conventional gas chromatography (1D-GC) and comprehensive two- dimensional gas chromatography (GCxGC) for the detailed analysis of petrochemical samples, Colombe Vendeuvre, Fabrice Bertoncini, Laurent Duval, Jean-Luc Duplan, Didier Thiébaut, Marie-Claire Hennion, Journal of Chromatography A, 2004
    >> https://doi.org/10.1016/j.chroma.2004.05.071
      
  5. [Lynch-2003] Physico-Chemical Analysis of Industrial Catalysts. A Practical Guide to Characterisation, John Lynch, 2003, éditions Technip
    >> http://www.editionstechnip.com/fr/catalogue-detail/735/physico-chemical-analysis-of-industrial-catalysts.html

links list

  • Print page

Follow IFP Energies nouvelles (IFPEN)

on Twitter

on YouTube