Hi Fred
I'm assuming you're referring to the use of the "individual
background" algorithm. I'm not sure why ISODAT uses only a single
baseline point in its 'individual' algorithm. I've coded this both ways
in the software I've written for GC-IRMS data processing, and a linear
approximation (defining background points before and after the peak)
almost always results in better precision and accuracy.
The one exception is when your analyte is severely tailing . . . in
which case your end baseline point will be reflecting both the actual
baseline and some contribution from the the peak tail of your analyte,
and data quality suffers. The front baseline point is more
representative of the true baseline. This can also be an issue with
both the front and back baseline points if you've got a coelutions.
However, when these become issues, its usually a sign that you need to
improve your chromatography.
The idea behind the dynamic background algorithm described in Ricci,
1994, was that it was supposed to solve some of these problems by
identifying true background points in the chromatogram, independent of
the location of peaks. In practice, as Gerry points out, the dynamic
algorithm can be compromised by spikes in the baseline, and may also end
up sticking a baseline point(s) far away from the peak of interest. I
had no more luck than the ISODAT software in my own implementation, so
vigilance in inspecting the baseline is necessary.
- Gavin
> HI Fred,
>
> They have different options in Isodat (version dependent). The
> dynamic background will give a series of background lines with slopes,
> but can have problems picking the correct spots to anchor the
> lines--and doesn't like jumps (aka dilution turning on or off). No
> matter which baseline method you pick, you have to review every run to
> make sure it's not gone off the deep end.
>
> Good luck with Mion.
>
> take care,
>
> gerry
>
>
>
> On 6/16/2011 4:49 AM, Fred Martin Kaaby wrote:
>>
>> Dear IsoGeochem
>>
>>
>>
>> It has always puzzled me how Isodat calculates background/baseline on
>> GC-peaks. It uses the datapoints before peaks start and assumes the
>> background are constant throughout the peak. Why don’t they follow
>> the regular convention for drawing baselines as in ordinary GC, from
>> background before start, to background after end (A linear
>> approximation)?
>>
>>
>>
>> Thanks for any insights.
>>
>>
>>
>> Fred Martin Kaaby
>>
>> (Who’s still working on Mion Chrom…)
>>
>
--
Gavin Sacks
Assistant Professor of Enology
Dept. of Food Science & Technology
NYS Ag. Experiment Station
Cornell University
Geneva, NY 14456
315-787-2458 (w)
607-592-1504 (c)
|