> Thank you for answer, I really appreciated.
> However, if I delete all the .mat files, the spatially normalised images
> don't appear as been flipped (even if the functional volumes are resliced
> or not).
> We would very much appreciate if you would give us a solution to this
> problem.
>
> Moreover, could you please better precise why should we do the step " Use
> the reslice only option of Realign or Coregister, to reslice all the
> volumes to match the copy "? I don't see very well the utility of this
> operation.

Trust me.  It should work.

Realign and coregister both include the option of reslicing images, without
doing any estimation procedure.  The act of reslicing means that afterwards,
images that are related to each other by .mat files become related so that
each voxel in one image file corresponds to each voxel in the other.  The
images are resampled so that the voxels match.

Flipping an image by its .mat file, and then reslicing it to an un-flipped
image means that the actual voxels get flipped.  If you want flipped and
unflipped images, then you can create mat files that say the images are
 flipped. These files can then be resliced to create images where the voxels
 are flipped. Deleting the original .mat files will then unflip the original
 images.

Best regards,
-John

> > > Would you be so kind as to help us with the following question related
> > > to SPM. We are interested in inter-hemispheric asymmetry of some
> > > specific visual processes. To assess the asymmetry of the BOLD
> > > responses, we have perfomed the following steps.
> > >
> > > 1. We have rendered symmetric the T1-template by averaging the original
> > > template and its "flipped" version. We checked that this was properly
> > > done by calculating in an ensuring step the "flipped" version of the
> > > "symmetrical averaged template", and by substracting the former from
> > > the latter.
> >
> > If you have enabled the "brain masking", whereby the registration is
> > weighted by some image (normally apriori/brainmask.img), then ensure that
> > this is also symmetric.
> >
> > > 2. We normalized our functional and anatomical data with respect to
> > > this symmetrical template, once according to the neurological
> > > convention, a second time according to the radiological convention.
> > >
> > > 3. We calculated the contrasts for both cases, excepting to obtain
> > > strictly symmetrical results in view of  contrasting the latter in an
> > > ensuring step. In pratice, we got nearly symmetrical results as may be
> > > judged on the "glass brains" attached.
> > >
> > > We would very much appreciate if you could tell us whether these slight
> > > asymmetries must be considered normal, or whether we are doing
> > > something wrong.
> >
> > I would expect there to be slight assymetries because not every voxel in
> > the template image is use in the estimation of the warps.  The template
> > is sampled every few voxels, so there may be some assymmetry in the way
> > this is done.
> >
> > It is more conventional to spatially normalise to the symmetric
> > templates, and
> > then flip your spatially normalised images.  An easy way of doing this
> > is:
> >
> > 1) Make a copy of one of your spatially normalised images.
> > 2) Use <Display>, setting a zoom in the x direction of -1, and applying
> > this zoom to all your spatially normalised images (except the copy).
> > This will create a .mat file for all the volumes.
> > 3) Use the reslice only option of Realign or Coregister, to reslice all
> > the volumes to match the copy.
> > 4) Delete all the .mat files.

--
Dr John Ashburner.
Functional Imaging Lab., 12 Queen Square, London WC1N 3BG, UK.
tel: +44 (0)20 78337491  or  +44 (0)20 78373611 x4381
fax: +44 (0)20 78131420  http://www.fil.ion.ucl.ac.uk/~john