Baseline fit algorithm

[sana_sankar]

I am trying to develop a baseline fit algorithm since last august. All attempts have failed. My baseline is simply going over the signal. I use spline fit for constructing the baseline.

Can anyone suggest a better method?

[tom jupille]

If you have insufficient information, you have insufficient information and any resulting quantitation is little better than guesswork. The only way to improve the situation is to improve the chromatography or the sample cleanup.

I'm not trying to be sarcastic here. You can't compute your way around inadequate chromatography, so you may as well do simple linear interpolation between key points.

[eselmeister]

Hi sana_sankar,

are you using a mass spectrometer?
If yes, you could have a look at the source code of the simple baseline detector I've developed for OpenChrom.
But seriously, it is very simple and shows some inaccuracies:

http://sourceforge.net/projects/baselinedetec


Simple baseline detection using OpenChrom


Further on, I do agree with tom, that it's necessary to pre-optimize the chromatography or to do a sample cleanup.
Anyhow, it only works if you are able to measure the samples again.
In other cases, a post-optimization is inevitable. That's why I've implemented a feature to include "Filters" in OpenChrom.
The latest filter is a denoising filter, to reduce chemical noise (a paper is on its way):


Baseline after applying the chemical noise reduction filter


It hopefully helps you and gives you inspiration.
There is also a paper I recommend:

SUN, X. Y.; SINGH, H.; MILLIER, B.; WARREN, C. H. & AUE, W. A.
Noise, Filters And Detection Limits
Journal Of Chromatography A, 1994, 687, 259-281



Cheers,
eselmeister