Just to add to Christian's response: You will probably have to left-right flip the ROI_MNI_V4.
Cheers-
Andreas
________________________________
Von: FSL - FMRIB's Software Library im Auftrag von Christian Beckmann
Gesendet: So 23.04.2006 02:36
An: [log in to unmask]
Betreff: Re: [FSL] Anatomical Automatic Labeling or something like this
Hi Michal,
you can use avwmaths to extrcats individual regions from e.g.
ROI_MNI_V4 (from the SPM toolbox) and then use these masks in
Featquery to calculate summary stats. Any ROI in MNI space should
also work.
More importantly, however, there seems to be something wrong with
your Z-stats images if the cluster size is >20000 voxels (unless your
voxel size is very small this basically implies whole-head
activation). I suspect that there is some structured noise effect
which is partially correlated with your GLM design. IMHO the things
to check are (i) does the histogram of Z-stats looks sensible (you
can do this fslview), (ii) is there some structured noise effect in
the data (maybe use melodic find these and to denoise the data;) and
(iii) try mixture modelling on the z-maps to see if that improves
things (see http://www.fmrib.ox.ac.uk/fsl/mm/index.html)
hope this helps
Christian
On 23 Apr 2006, at 00:53, Michal Kuniecki wrote:
> Hello everyon
>
>
>
> I have a problem, typical one I guess, however I was
> not able to find an answer on FSL list. I have got a very large
> activation (because I have a very large sample of 40 subjects)
> which goes in Z up to 22. If, using easythresh, I cut activations
> below Z=3.1 than I get only one cluster consisting of 26346 voxels
> and spanning through the entire brain. This is quite useless I
> think. Is there any way of getting some valuable information out of
> this? Namely SPM users have nice Anatomical Automatic Labeling
> program (http://www.cyceron.fr/freeware/) which is capable of
> spitting the statistics about how much of the given brain region is
> being covered by the activation (the template with names of brain
> regions is taken from ALL map provided with MRIcro). Is there
> anything comparable available for FSL results. If not how shell I
> report my data having one huge significant blob ? Rising the Z
> threshold is not an option as some, otherwise significant regions,
> would disappear, besides what rationale I would give to such an
> operation? That I was unable to accurately analyze my data at the
> p=0.001 ? Going through individual ROIs would be on the other hand
> extremely tedious.
>
>
>
> I would appreciate your help very much
>
>
>
> Best regards
>
> Michal Kuniecki PhD
>
> Department of Psychophysiology
>
> Institute of Psychology
>
> Jagiellonian University
>
> Al. Mickiewicza 3
>
> 31-120 Krakow, POLAND
>
>
> --
> No virus found in this incoming message.
> Checked by AVG Free Edition.
> Version: 7.1.385 / Virus Database: 268.4.4/319 - Release Date:
> 2006-04-19
>
>
>
--
Christian F. Beckmann
Oxford University Centre for Functional
Magnetic Resonance Imaging of the Brain,
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
Email: [log in to unmask] - http://www.fmrib.ox.ac.uk/~beckmann/
Phone: +44(0)1865 222551 Fax: +44(0)1865 222717
|