Baseline Correction in Arrayed Experiments

[GB, author of iNMR]

There is a drastic difference in the way iNMR applies phase correction and baseline correction. This behavior is peculiar of iNMR (and SwaN-MR before it) and, although people have been using these programs for twenty years, it's easy to forget. Actually, most people even confuse the two things. Any spectrum may require several kinds of corrections: a shift in time domain, a replacement of the first few points of the FID, drift correction, calibration of the ppm scale, etc… Phase and baseline corrections are just the two most frequently required corrections and they are completely unrelated to each other, though many students, new to NMR, collectively call them "phase correction".

The NMR spectrum is collected in two channels. Most of the times, neither of them is in phase with the signal. If, for example, the channel is in anti-phase, the signal is negative. To obtain positive signals it is necessary to mix the right amounts of the two acquired channels. This is best done after the acquisition, because it's not easy to predict the phase before performing the NMR experiment.

When this correction is done and all the peaks are positive, the spectrum may still contain an undesired component. If you integrate the regions, it may happen that even empty regions have non-zero integrals. In such a situation, the integrals of the signals may not correspond to the right values. The undesired component that pollutes the integrals is called "baseline". It's what remains of the spectrum if you remove all the signals. What we want to do, instead, is to remove the unwanted component and keep the signals. This procedure is called "baseline correction".

When working with a 2-D matrix, you can extract a single row or column and process it separately. After you close the extract, iNMR automatically copies the phase correction into the whole matrix, but it does nothing for the baseline correction! It's easy to understand why: each row/column of the matrix requires its own baseline correction, therefore it is quite wrong to apply the same unique correction (calculated on a single row/column) to the whole matrix. The results would be highly unpleasing to the eye. Fortunately there is no such problem for phase correction (the extremely rare exception was treated here: http://www.inmr.net/articles/phrow.html).

This is a trap a few users fall into. They erroneously believe that both phase and baseline corrections are magically copied into the whole matrix. Performing the correction on the individual row is harmless, of course. The problem is just that the correction disappears as soon as you close the extract and you must be aware of this fact.

THE CONCEPT TO REMEMBER IS: - When working with a matrix (for example a relaxation or kinetic experiment), perform the baseline correction on the whole matrix and not on the individual extracts.