Hi,
The full range of multiple-comparison options is currently being
implemented (since spatial relations are quite different and so
clustering is non-trivial), but with the current distribution you can
convert to and from an image format and do FDR or Bonferroni-style
correction. See below for some commands I've used to do
FDR correction.
All the best,
Mark
run_mesh_utils -m ${infile}.vtk -o ${infile}_Fvals --doScalarsToVolume
dof=`cat ${txtfile} | grep -i pillai | head -1 | awk '{ print $5 }'`;
dof2=`cat ${txtfile} | grep -i pillai | head -1 | awk '{ print $6 }'`;
echo "Degrees of freedom are: $dof $dof2"
run_mesh_utils -m ${infile}.vtk -o ${infile}_pvals.vtk --doFtoP --dof=
$dof --dof2=$dof2
run_mesh_utils -m ${infile}_pvals.vtk -o ${infile}_pvals --
doScalarsToVolume
thr=`fdr -i ${infile}_pvals -q 0.05 | grep -v '^ *[A-Za-z]'`;
echo "FDR threshold is $thr"
fslmaths ${infile}_pvals -uthr $thr -bin ${infile}_fdrmask
fslmaths ${infile}_pvals -mas ${infile}_fdrmask ${infile}_pthresh
fslmaths ${infile}_Fvals -mas ${infile}_fdrmask ${infile}_Fthresh
run_mesh_utils -m ${infile}.vtk -i ${infile}_Fthresh -o $
{infile}_Fthresh.vtk --doVolumeToScalars
On 23 Dec 2008, at 18:01, j janssen wrote:
> Hi,
>
> yes, thanks.
>
> doing:
>
> run_mesh_utils -m vertex_analysis_output.vtk -o vertex_pvalues.vtk --
> doFtoP --dof=X --dof2=Y
>
> will transform F to p. p may have to be corrected for multiple
> comparisons, can this be done?
>
> thanks,
> -joost
>
>
>
> On Tue, Dec 23, 2008 at 11:36 AM, Mark Jenkinson
> <[log in to unmask]> wrote:
> Hi,
>
> I think the version number probably didn't get updated in the source
> - sorry.
> As for the output, you need to turn the verbose flag on with -v.
>
> That should fix it.
>
> All the best,
> Mark
>
>
>
>
> On 23 Dec 2008, at 10:30, j janssen wrote:
>
> Hi,
>
> i installed this file on my 64 bit fedora core 7 linux:
> -rw-rw-r-- 1 neuro neuro 1370478353 2008-12-10 19:02 fsl-4.1.2-
> centos5_64.tar.gz
>
> when using the following command:
> first_utils --usebvars --vertexAnalysis -i concat_bvar-file -d
> design.mat -o output_mesh --useRigidAlign --useReconMNI --useWilks
>
> produces an on-screen output:
> model loaded
> done recon and reg
> new design matrix
> 1 1.23457 -0.419753 etc. etc.
>
> first_utils and first commands both give "... (Version 1.0)
> University of Oxford (Brian Patenaude)" while the FSL website states
> "version 1.1".
>
> the first_utils script states:
> cout<<"Wilk's "<<wilk<<" "<<F<<" "<<df1<<" "<<df2<<endl;
>
> and
>
> cout<<"Pillai F "<<pillai<<" "<<F<<" "<<df1<<" "<<df2<<endl;
>
> but i do not see this output anywhere. what did i do wrong?
>
> thanks,
> -joost
>
>
> On Mon, Dec 22, 2008 at 10:45 PM, Mark Jenkinson
> <[log in to unmask]> wrote:
> Hi,
>
> We have a new tool that will let you do this.
> You can get a beta version from:
> http://fsl.fmrib.ox.ac.uk/fsldownloads/patches/
> The relevant files are:
> fsl-macosx-runmeshutils.tar.gz
> or similar for other platforms.
>
> To install it just do:
> tar -zxvf fsl-macosx-runmeshutils.tar.gz
> mv run_mesh_utils $FSLDIR/bin
>
> Once you or your system admin has installed this
> program (run_mesh_utils) then you just need to do:
>
> run_mesh_utils -m vertex_analysis_output.vtk -o vertex_pvalues.vtk --
> doFtoP --dof=X --dof2=Y
>
> where X and Y are values that you can get from the
> original vertex analysis output (X and Y are the last
> two numbers of any line starting "Pillai F").
>
> You can then open the output (vertex_pvalues.vtk) in
> FSLView as with the other vertex analysis output vtk file.
>
> All the best,
> Mark
>
>
>
>
>
>
> On 18 Dec 2008, at 21:10, Pinghong Yeh wrote:
>
> Dear all,
>
> For the vertex analysis, I wonder how to output the p-values of F-
> statistic
> after running first_utils on the FIRST results? Thanks.
>
> Pinghong
>
>
>
