Dear Anne-Sofie,

Thank you for the upload.
This file is indeed worrying, as not only are the F-values negative, 
which I have never seen, but they are also very far from zero.
I am assuming that you have checked the registration and segmentation
results for the individual subjects and are satisfied with them.

It is possible that your first_utils command line was incorrect.
Note that the parts of the command that are written like [ xxx ]
are actually optional parts and you should not include them all.

MNI space is a standard space (an average of many individuals)
and native space is the space of the subject without adjusting it
to fit the average.  The main difference is that if your subjects are
different from the "average" (e.g. older) then there can be a difference
between the MNI space and native space results (as the former will
scale each image to match the average/standard space).  I would
recommend the MNI space analysis in general (as it adjusts for
head size).  So just use the --useReconMNI option.

Also, it is important what your design matrix looks like.

So try to do the analysis again with the --useReconMNI option.
If the results are still odd (either negative F-values or very non-zero)
then please email us back with the exact first_utils command line 
that you used plus the design matrix and vtk file (use the upload site).
Hopefully this won't be necessary and you will get some sensible
results from this analysis.

All the best,
	Mark





On 30 May 2011, at 13:40, SERMIIIN Dz wrote:

> Maybe I choose the wrong file, but i tried uploading it again and the number is 624641
> This is the output file of our vertex analysis before we run surface_fdr
> I actually think we used both --useReconMNI or --useReconNative  which of course is wrong!! But I'm not sure what the difference are and which one I should use.. I don't know what a native space is or a MNI space. 
> Thank you for our help so far!
> Anne-Sofie
> 
> 
> 
> 
> > Date: Mon, 30 May 2011 13:17:15 +0100
> > From: [log in to unmask]
> > Subject: Re: [FSL] vertex analysis F-values
> > To: [log in to unmask]
> > 
> > Dear Anne-Sofie,
> > 
> > You haven't uploaded an F-statistic vtk file - you've sent
> > me a p-value one. It is also very unusual in that all the
> > p-values are set to 1.0. I think something has gone wrong.
> > Try uploading the F-statistic output (directly from first_utils
> > output and not after surface_fdr).
> > 
> > You also should not use the -n option in the vertex analysis
> > without using the --usePCAfilter option, and we do not 
> > recommend that option anyway. Also, are you using 
> > --useReconMNI or --useReconNative ?
> > 
> > All the best,
> > Mark
> > 
> > 
> > On 30 May 2011, at 13:07, SERMIIIN Dz wrote:
> > 
> > > First we used run_first_all to segment the structures
> > > Then we concated all the bvars files using concat_bvars in the same sequence that we used to build the design matrix.
> > > then we ran the vertex analysis first_utils --vertexAnalysis --usebvars -i concatenated_bvars -d design.mat -o output_basename [--useReconNative --useRigidAlign ] [--useReconMNI] [--usePCAfilter -n number_of_modes] with number of modes 30 for hippocampus and caudate, and 40 for putamen (we are only interested in these three structures). 
> > > We open e.g. (name).bvars1.vtk which is the analysis with respect to development in reaction time. (because it is what we put in the first column of the design matrix). I uploaded a example with reference nummer 756006
> > > after that we used the function surface_fdr, where we think the results look ok, but we want to make sure that the first part was performed correctly, because of the negative f-values.
> > > Anne-Sofie
> > > 
> > > 
> > > 
> > > > Date: Mon, 30 May 2011 12:29:32 +0100
> > > > From: [log in to unmask]
> > > > Subject: Re: [FSL] vertex analysis F-values
> > > > To: [log in to unmask]
> > > > 
> > > > Dear Anne-Sofie,
> > > > 
> > > > There should not be any negative F-values. Are you sure you
> > > > are looking at the correct output? It would also be useful to
> > > > have a copy of the exact commands you have been running
> > > > (just for one of the images will be fine).
> > > > 
> > > > You cannot attach files to the email list (unless they are very
> > > > tiny) so instead please upload the file to:
> > > > http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi
> > > > and send us the reference number.
> > > > 
> > > > All the best,
> > > > Mark
> > > > 
> > > > 
> > > > 
> > > > On 30 May 2011, at 11:58, SERMIIIN Dz wrote:
> > > > 
> > > > > 
> > > > > 
> > > > > I have a question about the vertex analysis. The result is a vtk-file including the F-statistic value. My partner and I are writing a bachelor project with 60 t1 MR-scans, segmented the subcortical structures using FIRST and ran the vertex analysis. Our F-values are negative, and the F-distribution says that F-values will never be negative. How can that be?? We are running the analysis with respect to reactiontime, gender and testing two groups against each other. Almost all the f.values in the vtk output files are negative. I added a vtkfile so you can see it yourself. We ran the test with number of modes 30 for hippocampus, 30 for caudate and 40 for the putamen. Does this number means a lot in the analysis? We tried 20 for all the structures first, but the results weren't the same, compared to them with 30/40. We used 30/40 according to the table in the report for the program.
> > > > > 
> > > > > I hope you a able to help us, so we are sure that what we did was right.
> > > > > Anne-Sofie Lindberg
> > > > >