Login  Register

Re: Peak fitting with ImageJ

Posted by David Hovis on Feb 03, 2008; 7:23pm
URL: http://imagej.273.s1.nabble.com/Label-images-tp3697295p3697300.html

Wolfgang,

I'm not really familiar with the gel tools, but I think what you are  
asking is not trivial.  I use fityk (http://fityk.sf.net/) for general  
1-D peak fitting.  Fityk is free and comes in GUI, CLI, and library  
versions.  I've hacked a JNI interface with SWIG to use libfityk  
directly with my custom ImageJ plugins.  It is very fast, but it is  
decidedly non-trivial.

Often when there are overlapping peaks, you need to take care with  
your model to make sure that you get a physical result.  Oftentimes,  
closely spaced peaks can be fit reasonably well with a single wider  
peak.  If you know that there is some reasonable maximum width to your  
peaks, then there are ways of applying constraints in fityk to prevent  
the fitting algorithm from converging on non-physical results.

--David

----------------------------------------
David Hovis
Senior Research Associate
Department of Materials Science
Case Western Reserve University
[hidden email]

On Feb 3, 2008, at 8:55 AM, Wolfgang Schechinger wrote:

> Dear ImageJ experts,
>
> is there a way to do peak fitting within ImageJ? I habe obtained  
> some chromatogram-like curves with the Crtl+1,2,3 sequence and would  
> like to estimate if there is more than one signal hidden in one peak  
> (i.e. if one peak may be expressed as the sum of two, partially  
> conincidental, peaks).
> Alternatively, is there a way to convert and save these curves as  
> numbers in e.g. a csv or text file that one may process with some  
> third party software (any recommendation for which software might be  
> useful are highly appreciated as well)
>
> Many thanks!
>
> Wolfgang
> --
> Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
> Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer