[Brains-users] Problems with tissue class models and linux paths

[Ronald Pierson]

In the output when running segmentation look for the lines that start
with "Image".  This is output after each iteration of plug picking. 
This shows the average intensities of the tissue classes for each
image.  Image 0 is the T1, and the classes listed in the permutations
are 1 - gray, 2-white, 3-CSF.  The numbers need to make sense, and NOT
be zero or all the same.  For instance, here is one example:

Image 0:   Centers:  118.554   95.155   32.262
Image 0:   Permutation:  2   1   3
Image 1:   Centers:  67.059   96.564   247.452
Image 1:   Permutation:  2   1   3
Image 2:   Centers:  113.759   153.807   215.440
Image 2:   Permutation:  2   1   3

Can you cut-and-past these lines from your segmentation output (or all
of it would be fine, too), and we can see what is going on?

Has anyone been able to run segmentation with CygWin version?

Ron

On Mon, 2005-09-12 at 14:27, Ronald Pierson wrote:
> This sounds like the most likely.  Load the masks csf_plugs.mask,
> grey_plugs.mask and white_plugs.mask and measure them.  If one of them
> is zero, that is likely the case.  Often when there is a problem both
> the grey and csf plugs are empty.  
> 
> This can often be due to the scans being too dark (check the T1
> especially), or if the sujects are quite young and have very little csf,
> or if the scan is very noisy.  
> 
> First check the T1, if it is dark try using Utilities, Image Tools,
> Image math to add the T1 to itself, then save it.  If this doesn't
> brighten it up (you will see it on the screen) there may be some bright
> spots in the scan that are causing the rest of the brain to have a
> pretty small part of the dynamic range.  I have a routine or two that my
> help you if that is the case.
> 
> The other fixes depend on which of these problems is present.  I'll try
> to put together something more comprehensive in the next day or two.
> 
> Ronald Pierson
> 
> 
> On Mon, 2005-09-12 at 12:04, Greg Harris wrote:
> > Kevin,
> > 
> > Note that it will "fail to create the model" if any of the
> > training class plug masks has zero volume (is empty).
> > In that case, it's the plug-picking parameters that
> > need special attention.
> > 
> > Greg Harris
> > 
> > 
> > Michael Kiguta wrote:
> > 
> > > Hi Kevin,
> > >
> > > I think what you need to do is save the segmentation script then open 
> > > it as a nedit file. This enables you to examine all the parameters 
> > > used in the tissue classification as well as the various file paths 
> > > where the plugs as well as masks are stored. Examine each file path to 
> > > ensure it is correct. The program will save and find those masks where 
> > > you've specified they should be. I hope that helps.  Any other ideas 
> > > Helen?
> > >
> > > ../Mike
> > >
> > >>
> > >> Kevin Berne wrote:
> > >>
> > >>> Hello,
> > >>> After my request last week I have installed the recommended Brains2 
> > >>> DEBUG 2005_09_05 as well as an up-to-date Cygwin environment. This 
> > >>> resulted in a much smoother operation without the troublesome 
> > >>> inability to create a standard directory. Nevertheless I have 
> > >>> encountered more difficulties, which as a linux novice I have tried 
> > >>> but failed to solve alone.
> > >>> First of all, whenever I try to run a segmentation to create a 
> > >>> tissue model, the process fails after writing the plugs, displaying 
> > >>> an error message expressing either the inability of creating the 
> > >>> model or loading certain masks. This occurs in several forms 
> > >>> whatever options I seem to choose within the dialog boxes.
> > >>> Furthermore, one certain error message keeps popping up whether I 
> > >>> try to run a segmentation or a neural net, reading "Error: chgrp: 
> > >>> invalid group `stdwkup´. I have tried to conduct my own 
> > >>> internet-research, but since I do not have any experience with 
> > >>> linux-based operating systems I can not make sense of what I mostly 
> > >>> find.  I would gladly appreciate any recommendations or tips anyone 
> > >>> could give me about this particular malfunction.
> > >>> Thank you very much for your time,
> > >>> Kevin Berne
> > >>>
> > >>> __________________________________________________________________________ 
> > >>>
> > >>> Erweitern Sie FreeMail zu einem noch leistungsstärkeren 
> > >>> E-Mail-Postfach!       Mehr Infos unter 
> > >>> http://freemail.web.de/home/landingpad/?mc=021131
> > >>>
> > >>>
> > >>> _______________________________________________
> > >>> Brains-users mailing list
> > >>> Brains-users at www.psychiatry.uiowa.edu
> > >>> http://www.psychiatry.uiowa.edu/mailman/listinfo/brains-users
> > >>>
> > >>
> > >
> > >
> > 
> > _______________________________________________
> > Brains-users mailing list
> > Brains-users at www.psychiatry.uiowa.edu
> > http://www.psychiatry.uiowa.edu/mailman/listinfo/brains-users
> > 
> 
> _______________________________________________
> Brains-users mailing list
> Brains-users at www.psychiatry.uiowa.edu
> http://www.psychiatry.uiowa.edu/mailman/listinfo/brains-users
>