Dear Bjorn,
When using POSSUM you get an output directory in with a file called
brain.nii.gz . This file consists of three different volumes one for
each tissue type. This is the input to POSSUM and not a structural
image. It is a segmented digital brain which consists of three
volumes (one for gray, one for white and one for CSF). So if you are
using this file to segment it you would get the kind of errors you
are complaining of.
Similarly, when you want to simulate an image you do not use a
structural image as your input, but as explained above, a segmented
version of the structural (one volume per tissue type, all merged
together). It is important to keep in mind that the MR parameters
(file called MRpar) need to be in the same order as the tissue types.
If you are using MR parameters given by default with POSSUM, they are
ordered as: first row is grey matter, second is white, and the third
is CSF so try to keep the order of your segmented volumes the same.
Once you finish the simulation this input file will be present in
your output directory and will be called brain.nii.gz.
The actual output of POSSUM is called image_abs.nii.gz and this is
the image you should be segmenting. fslview always helps to see your
data so that you can see these things easier. In addition there is
a very straightforward explanation of the files on the website:
http://www.fmrib.ox.ac.uk/fsl/possum/index.html . I highly recommend
you to read this website.
Hope this helps,
Ivana
On 18 Mar 2008, at 10:32, Bjorn Roelstraete wrote:
> dear all,
>
> when i try to analyze my simulated possum data with SPM, i get
> stuck at the segmentation part. the default "structural" image in
> possum is brain.nii, but when i use this file as the structural
> image in SPM,i get errors ("Matrix is singular to working precision").
> And when i try to use another structural file in possum
> (e.g.structural_brain.nii), it won't simulate! how do i solve this?
> is there a file in fsl or fsl-feeds that is suited for simulating
> with possum AND analyzing with SPM at the same time?
>
> thanks in advance,
>
> --
> Bjorn Roelstraete
> Ghent University
> Department of Data-analysis
> H. Dunantlaan 1, B-9000 Gent, Belgium
> Tel: 32-9-2646434
> E-mail: [log in to unmask]
>
|