Automatic Concurrent Phase & Baseline Correction in 1D NMR Spectra

POSTER by 1Ester Maria Vasini, 2Carlos Cobas, 1Stanislav Sykora
1Extra Byte, Castano Primo, Italy
2Mestrelab Research, Santiago de Compostela, Spain

Presented at SMASH 2017, Baveno (Italy), September 17-20, 2017.

DOWNLOAD full poster: PDF DOI permalink: 10.3247/SL6Nmr17.004 Stan's Library | Stan's HUB

Please, cite this online document as:
Vasini E.M., Cobas C. and Sykora S.,
   Automatic Concurrent Phase & Baseline Correction in 1D NMR Spectra,
   Poster at SMASH 2017, Baveno (Italy), September 17-20, 2017, DOI: 10.3247/SL6Nmr17.004.

Abstract

The algorithmic problem of phase correction (PC), and that of baseline correction (BC), of 1D NMR spectra have been both tackled many times over the last half a century, by many authors.
 
There are many algorithms which emulate the manual procedure. Basically, they all 'fit' the parameters which describe the phase (ph0, ph1) and the baseline (various parametrized models) so as to maximize some quality assessment of the corrected spectrum. Historically, the employed 'quality functions' included peak heights, negative peak lobes, DISPA patterns symmetry, selected baseline points, peak ablation, etc.
 
Here we propose a radically different type of 'quality function' Q to be optimized. It is based on the histograms of the spectrum (real and imaginary parts) which turn out to be very sensitive to both phase and baseline distortions.
 
This permits us to:
1. Carry out the phase and baseline corrections simultaneously. In traditional approaches these are always intended as separate evaluation steps and carried out separately (first phase and then baseline), even though all NMR spectroscopists know that the two corrections interfere with other. We think that our approach neatly overcomes this problem.
2. Carry out both corrections on both the real and imaginary parts of a spectrum. So far, the baseline correction was always done only on the real part, a fact that can have various adverse effects on other evaluation tasks.
3. Enhance the objectivity of the corrections, especially considering that in practice one often faces situations with multiple acceptable 'solutions'.
 
We describe the algorithm we have developed and we illustrate the results achieved.
 

Stan's Library Stan's HUB
Copyright ©2017 Sykora S., Vasini E.M. Page design by Stan Sykora