Hi,
Thanks for uploading your data.
Unless you have a teeny, tiny brain then I think your
voxel size is wrong. Currently your image is
64 x 64 x 21 with 0.9375 x 0.9375 x 0.9375 mm
voxels. This is about a factor of 4 smaller than I'd
expect it to be for a human brain.
This also explains why neither mcflirt nor flirt work
properly, as both expect a human-sized brain and
have multi-scale analysis tailored for this, with 8mm,
4mm, 2mm and 1mm stages. With your image
presently, the 8mm version is only a couple of voxels
and hence not useful which will be why it is failing.
So I would check your reconstruction of this image.
All the best,
Mark
On 17 Feb 2010, at 16:38, Jennifer Evans wrote:
> Hi,
>
> This is a really basic question about getting mcflirt/flirt to do
> what I
> expect. I just wanted to do simple motion correction to one of my
> fMRI
> timeseries, but mcflirt seems to add "motion" rather than correcting
> for
> what little motion there is. I checked the data by running another
> motion
> correction algorithm and the data seem fine. I'm wondering whether
> the
> algorithm is being biased by the intensities in the ventricles... I
> just
> ran mcflirt -in <dataset>.
>
> This is only a problem because I would like to run multi session
> temporal
> concatenation in Melodic which requires a flirt registration step
> that does
> similar odd things to my data (like shrinking the brain so it's really
> small) and the ICA fails.
>
> Any insight into what might be the problem would be appreciated.
>
> Thanks,
> Jen
>
|