I really think with this data that estimating spatially normalisation
parameters from it is not going to work particularly well. If you
acquire other images of the same subjects, but without the surface coil, then
these could be used to estimate the warps (providing you can determine the
transformation between the surface coil images and the others.
I have been working on an intensity non-uniformity correction approach, but
it won't perform the miracles needed to get these images into a form that
will allow them to be spatially normalised. If anyone is interested, they
can pick up an early prototype from:
ftp://ftp.fil.ion.ucl.ac.uk/spm/flatten
It is a very early version and hasn't been extensively tested yet. As a
thoroughly tested alternative, you may wish to try N3, which is available
from http://www.bic.mni.mcgill.ca/software/N3/
Best regards,
-John
On Friday 02 November 2001 14:06, Heledd Hart wrote:
> > Hi there,
> I am trying to analyse sagitally-acquired surface coil data of the
> temporal lobes, the intensity of which fades out towards the midline,
> leaving me with data for only one hemisphere. I am trying to get the
> data from 10 subjects in a common space to do a group analysis and have
> encountered some problems.
>
> Since the data is only of a partial brain volume and normalising it to
> the whole brain SPM99 template doesn't work I have made a "template"
> using an average image from all the subjects to attempt to at least
> align data from the subjects into the same space. Co-registering the
> data from single subjects to my surface coil template works ok, but when
> i attempt normalisation to this same template SPM re-orients the images
> very strangely. Not being an expert by any stretch of the imagination I
> find it quite difficult to describe the way in which the data has been
> re-oriented, so have attached a document which shows a picture obtained
> using the Check Reg function in SPM and illustrates the template used
> and the way the data looks before and after normalisation.
>
> I have checked that the .mat files are the same for the data and the
> template prior to normalisation so don't think this is the problem. I
> think it must lie in one of the stages occuring in the normalisation
> process (that I don't know too much about) where SPM expects something
> to occur in the images which is somehow different for my surface coil
> data. Defaults I have tried are R=R (as this should be true for this
> data); disallow customised normalisation; 7x8x7 basis functions; 12
> non-linear iterations; medium regularisation; no brain mask; don't mask
> object; default bounding box and 3x3x3 voxel size (as this is the
> resolution of the data and the template).
>
> I'd be extremely grateful for any advice or suggestions anyone may have
> to offer on where I might be going wrong and, if possible, how to put it
> right.
>
> Thank you very much.
>
> Heledd Hart.
>
> MRC Institute of Hearing Research
> University Park
> Nottingham
> NG7 2RD.
>
> Tel: (0115) 9223431.
> Email: [log in to unmask]
----------------------------------------
Content-Type: application/msword; charset="us-ascii";
name="surfacecoilnormalisation.doc"
Content-Transfer-Encoding: base64
Content-Description:
----------------------------------------
--
Dr John Ashburner.
Wellcome Department of Cognitive Neurology.
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
mail: [log in to unmask]
|