Baseline Correction

[Richard Wallace Briggs]

Is the Baseline Correction feature not available in the preprocessing module? It is listed and described in the Help section of jMRUI 5.2, but is not shown in the pull-down Preprocessing menu or on the new Web page, that I can find.

Richard Wallace Briggs
Georgia State University

[Michal Jablonski]

Dear Richard,

I will check it and I will let you know.

Best Regards

[Miquel]

The Baseline Correction feature was available on jMRUI up to version 3.0 released on July 2006, and it was removed from the software before the release of version 4.0, which happened on September 2009.

From what I recall, there were several reasons that supported the removal of the baseline correction feature:

• the algorithm was a sort of hack, with the signal being processed in the frequency domain, and then converted back to the time domain, and last subtracted from the original spectrum;
• there were no studies or documented evidence of the algorithm reliability;
• the use of a smoothing algorithm to remove an arbitrary baseline was regarded as misleading and substandard, and it was feared it could yield distorted spectra and artefactual results.

Unfortunately, we forgot to remove the reference to this algorithm in the software help of that version and later releases (that’s called “error propagation” 😉

Miquel Cabanas, SeRMN
Universitat Autònoma de Barcelona

[Miquel]

I have marked in the description of the algorithm the points that worried us and that ultimately lead to the removal of the algorithm,

Baseline Correction – The baseline correction is another frequency domain operation in jMRUI [1]. It is a method to cosmetically correct for a background signal [2] With cosmetically we indicate that its physical relevance is limited, [3] but we have nevertheless been able to quantify signals after a baseline correction that were not easy to quantify before. The correction consists of the calculation of an average value for each point in the spectrum. [4] The number of points to be used for this average is to be entered [5]. This average is then subtracted from the original spectrum and the result is reverse Fourier Transformed to obtain a time domain signal [1].

[1] jMRUI aims at processing MRS signals in the time-domain to avoid the problems affecting Fourier Transformed signals in the frequency-domain.
[2] Cosmetic changes should change the aspect of the signal and not the signal itself, but in this case it actually changed the signal.
[3] If it has no physical relevance and/or interpretation, it should not be taken into account because there is a danger of misleading users & causing artefactual results.
[4] This is a smoothing technique which basically filters out the narrow components of the spectrum and preserves the broad ones. Thus, the subtracted baseline signal will contain information pertaining to the spectrum which should not be removed unless done on purpose.
[5] This is arbitrariness, and it was against one of the goals of the jMRUI software: to process MRS signals using physicochemical meaningful models.

Miquel Cabanas, SeRMN
Universitat Autònoma de Barcelona

[Author]

Posts