Auto alignment for quantitative pairwise differencing of 2D data (Daniel Geschwender, MDCW 2023)
- Photo: MDCW: Auto alignment for quantitative pairwise differencing of 2D data (Daniel Geschwender, 2023)
- Video: LabRulez: Daniel Geschwender: Auto alignment for quantitative pairwise differencing of 2D data (MDCW 2023)
- 🎤 Presenter: Daniel Geschwender¹, Qingping Tao¹, Chase Heble¹, Stephen Reichenbach¹,² (¹GC Image, LLC, Lincoln, USA. ²University of Nebraska, Lincoln, USA)
💡 Book in your calendar: 15th Multidimensional Chromatography Workshop (MDCW) January 2024
15th Multidimensional Chromatography (MDC) Workshop 2024
Abstract
Identifying chemical differences among samples is useful for process monitoring, sample classification or identification, correlative determinations, and other important tasks. Multi-dimensional chromatography (including GCxGC and LCxLC) is a powerful technique for highly effective chemical separations of complex mixtures, but it also produces highly complex data that require both interactive and automated comparative analysis methods.
One common case is to compare two types of samples to determine similarities and differences. Our existing side-by-side interactive differencing tool provides both comparative visualization of chromatograms and comprehensive characterization of sample differences down to individual peaks. It allows comparing chemical differences between two samples qualitatively and quantitatively. However, it requires that the analyst specify an initial peak correspondence in order to compute a transform from one retention time space to the other.
We propose a new automated matching algorithm that combines computer vision image registration techniques with the peak matching algorithm utilized in our Investigator framework. Computer vision-based registration provides an intuitive and unsupervised way to identify and match patterns across entire raw chromatograms. This registration computes an initial transform to seed our peak matching algorithm. The peak matching algorithm identifies and matches reliable peaks based on several criteria including retention time distance and spectral match scores. This matching is used to compute a refined transform that is then provided to the side-by-side interactive differencing tool in order to compare the chromatograms. We demonstrate the technique to compare across different samples as well as across different detectors using the same sample.