Hi Rosalia,
You can compute the spatial correlation within a user-defined mask using
the following (in c-shell syntax):
######### Pearson's R correlation between two image volumes
#############
## --- Compute the correlation from 1st principles
## --- i.e., account for the mask when determining which voxels to
include
## --- in the computation
# Note that within the mask, we will treat 0's as valid data
set M1 = `fslstats $img1 -k $mask -m`
set M2 = `fslstats $img2 -k $mask -m`
fslmaths $img1 -sub $M1 -mas $mask demeaned1 -odt float
fslmaths $img2 -sub $M2 -mas $mask demeaned2 -odt float
fslmaths demeaned1 -mul demeaned2 demeaned_prod
set num = `fslstats demeaned_prod -k $mask -m`
fslmaths demeaned1 -sqr demeaned1sqr
fslmaths demeaned2 -sqr demeaned2sqr
set den1 = `fslstats demeaned1sqr -k $mask -m`
set den2 = `fslstats demeaned2sqr -k $mask -m`
set denprod = `echo "scale=4; sqrt($den1*$den2)" | bc -l`
# The mean can be used instead of the sum because the
# factor N/sqrt(N*N) will cancel
set true_r = `echo "scale=4; $num/$denprod" | bc -l`
-----
Or, AFNI has a function called '3ddot' that allows one to specify a mask
file. (Be sure to use its -demean option if you want the correlation).
Also, '3ddot' allows one to compute the eta^2 between two images if you
are interested in that measure of spatial correspondence.
cheers,
-MH
On Thu, 2012-02-02 at 16:10 +0100, Rosalia Dacosta wrote:
> Dear Michael,
>
> I am using fslcc command too for making my correlations between
> melodic_IC.nii.gz and metaICA_4mm.nii.gz template for identifying
> anatomical and psychological components, discarding
> artifacts ...so....in your opinion, what would be the best option?
>
> Cheers,
>
> Rosalia.
>
> 2012/2/2 Michael Harms <[log in to unmask]>
> Hi David,
> Note that fslcc includes ALL voxels in its calculation of the
> spatial
> "correlation" including voxels with 0's outside of the brain.
> In my
> opinion, this is not a meaningful estimate of the true spatial
> correlation.
>
> cheers,
> -MH
>
> On Thu, 2012-02-02 at 07:40 +0000, Stephen Smith wrote:
> > Hi - I'm not sure which analysis you're referrring to, but
> for simply
> > matching up maps (as opposed, for example, to doing a
> dualregression),
> > yes we would probably just use fslcc
> >
> >
> > Cheers.
> >
> >
> >
> >
> >
> > On 1 Feb 2012, at 18:33, David V. Smith wrote:
> >
> > > OK, great -- thanks for both sets of images! Can you also
> verify
> > > that one of the methods below is the way you guys perform
> the
> > > spatial cross-correlation?
> > >
> > > fsl_glm -i IC01.nii.gz -d PNAS_Smith09_rsn10_3mm.nii.gz -o
> PE --
> > > out_z=z_score -m mask.nii.gz
> > > where IC01.nii.gz is just the first image in the
> melodic_IC image
> > > (fslroi melodic_IC.nii.gz IC01 0 1)
> > >
> > >
> > > And then report the PEs as the correlation estimates? This
> method
> > > seems odd because of the extremely high z-scores and PEs
> aren't
> > > explicitly r values for a correlation...
> > > [smith@node52 groupICA_AU]$ more PE
> > > 0.24856 0.437975 -0.135177 -0.0728717 0.0331276
> -0.0438093
> > > -0.0734834 0.0104588 0.0270564 0.0727842
> > > [smith@node52 groupICA_AU]$ more z_score
> > > 91.2948 141.214 -41.624 -25.2608 10.92 -13.4459
> -20.1861
> > > 3.80642 9.4519 24.6514
> > >
> > >
> > > Or did you guys use fslcc?
> > > fslcc RSN02.nii.gz IC01.nii.gz (in the 3rd column, this
> has 0.46 as
> > > the correlation)
> > > where RSN02.nii.gz is the second image in the RSN template
> image
> > > (fslroi PNAS_Smith09_rsn10_3mm.nii.gz RSN02 1 1)
> > >
> > >
> > > If both methods are valid, is one preferred over the
> other?
> > >
> > >
> > > Thanks again!
> > >
> > >
> > > Cheers,
> > > David
> > >
> > >
> > >
> > >
> > >
> > >
> > > On Feb 1, 2012, at 12:02 PM, Stephen Smith wrote:
> > >
> > > > Hi - Christian has put this online now:
> > > > http://www.fmrib.ox.ac.uk/analysis/royalsoc8/
> > > > Cheers.
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > On 1 Feb 2012, at 08:26, Stephen Smith wrote:
> > > >
> > > > > Hi, the RSN/BrainMap images are at
> > > > > http://www.fmrib.ox.ac.uk/analysis/brainmap+rsns/
> > > > > We were going to put the Beckmann 05 images on there
> as well -
> > > > > haven't gotten around to that yet…
> > > > > Cheers
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > On 1 Feb 2012, at 06:58, David V. Smith wrote:
> > > > >
> > > > > > Hello,
> > > > > >
> > > > > > Did these RSN template maps (from the 2005 Beckmann
> paper?)
> > > > > > ever get posted to the web? I can't find any other
> mention of
> > > > > > it on the forums, and I'm not exactly sure where to
> look on
> > > > > > the FMRIB site.
> > > > > >
> > > > > >
> > > > > > Thanks!
> > > > > > David
> > > > > >
> > > > > >
> > > > > >
> > > > > > On Sep 6, 2011, at 12:13 PM, Stephen Smith wrote:
> > > > > >
> > > > > > > Hi - we've been meaning to put the brainmap/RSN
> PNAS maps up
> > > > > > > on the web for a while but keep not getting around
> to it -
> > > > > > > we'll try to get these up by the end of September.
> > > > > > > Cheers.
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > > On 2 Sep 2011, at 06:37, Eva Kenny wrote:
> > > > > > >
> > > > > > > > Hi,
> > > > > > > > I was planning on using the template matching
> procedure on
> > > > > > > > my Melodic data to identify ICs of interest and
> wondered
> > > > > > > > if there were template RSNs available?
> > > > > > > > Many thanks,
> > > > > > > > Eva
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> ---------------------------------------------------------------------------
> > > > > > > Stephen M. Smith, Professor of Biomedical
> Engineering
> > > > > > > Associate Director, Oxford University FMRIB
> Centre
> > > > > > >
> > > > > > > FMRIB, JR Hospital, Headington, Oxford OX3 9DU,
> UK
> > > > > > > +44 (0) 1865 222726 (fax 222717)
> > > > > > > [log in to unmask]
> http://www.fmrib.ox.ac.uk/~steve
> > > > > > >
> ---------------------------------------------------------------------------
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > > >
> ---------------------------------------------------------------------------
> > > > > Stephen M. Smith, Professor of Biomedical Engineering
> > > > > Associate Director, Oxford University FMRIB Centre
> > > > >
> > > > > FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> > > > > +44 (0) 1865 222726 (fax 222717)
> > > > > [log in to unmask]
> http://www.fmrib.ox.ac.uk/~steve
> > > > >
> ---------------------------------------------------------------------------
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > >
> > > >
> > > >
> ---------------------------------------------------------------------------
> > > > Stephen M. Smith, Professor of Biomedical Engineering
> > > > Associate Director, Oxford University FMRIB Centre
> > > >
> > > > FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> > > > +44 (0) 1865 222726 (fax 222717)
> > > > [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> > > >
> ---------------------------------------------------------------------------
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > >
> > >
> >
> >
> >
> ---------------------------------------------------------------------------
> > Stephen M. Smith, Professor of Biomedical Engineering
> > Associate Director, Oxford University FMRIB Centre
> >
> > FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> > +44 (0) 1865 222726 (fax 222717)
> > [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> >
> ---------------------------------------------------------------------------
> >
> >
> >
> >
> >
> >
> >
>
>
|