Hi - there's a couple of things you should be doing differently,
however neither (I think) explains the crashing - however, change
these things and see if it then runs:
- it is easier to use 1-column format for your confounds - full of
0s and 1s - that way you don't need to worry about getting the timing
exactly right. So each row is the value for one timepoint.
- you should create one new confound EV _separately_ for every bad
timepoint.... because each one will have a different 'bad' intensity,
so needs separate modelling.
You can achieve both points above via a single confound file, simply
by using the "Add additional confound EVs" text file option with
Nspikes columns and Ntimepoints rows.
Cheers.
On 10 Jun 2009, at 20:53, Emma Liang wrote:
> Hi Dear Steve,
>
> I was trying to add a confound EV. I use 3 columns files, and use
> cat -v check, no hidden character; use wc check, the lines of
> confound EV equal to the sum of the 8 original EV lines. (I have 7
> task EVs and the 8th EV for scanner spikes). In the confound EV text
> file, all the spikes are set to 1 in 3rd column and other tasks are
> to 0 respectively. But the full model setting always prompt
> 'terminate called after throwing an instance of NEWMAT: Incompatible
> Dimension Exception'. Is there any mistake on my EV setting? I'll
> appreciate it if you have time to give a look at my timing files
> attached. The timing is from log file of Presentation(NBS).
>
> Thanks,
> Emma
>
> On Wed, Jun 10, 2009 at 3:23 PM, Steve Smith <[log in to unmask]>
> wrote:
> Hi - it depends how the artefact is interacting with the signal of
> interest. If it's very strong then if may be best to ignore that
> timepoint completely, in which case you could add an extra regressor
> into FEAT that is zero at all timepoints except when the spike occurs.
>
> Cheers.
>
>
>
>
> On 9 Jun 2009, at 16:02, Emma Liang wrote:
>
> Dear Dr Smith,
>
> Thanks a lot. The ICA paper is very helpful. But I am a very beginer
> on signal processing methods, still need time to digest it. My
> current comprehension is that each MELODIC component is a part of
> original signal throughout the whole time serial. In my data, the
> scanner 'spike' noises affected each component and made the real
> activation signal very insignificant. I want to filter the EPI
> imaging by cutting all the intensity over 1000(machine spikes at
> over 2000) to 1000 and make them as a task. Is it ok? and is there a
> function to do that? Sorry if the question is too silly.
>
> Cheers,
> Emma
>
> On Fri, Jun 5, 2009 at 4:07 PM, Steve Smith <[log in to unmask]>
> wrote:
> Hi, see the FAQ:
>
> http://www.fmrib.ox.ac.uk/fslfaq/#melodic_filter
>
> Cheers.
>
>
>
> On 3 Jun 2009, at 12:42, Emma Liang wrote:
>
> Dear FSLers,
>
> I have a EPI imaging with some white noise slices throughout(about
> 20 of 500 volumes contaminated) because a unknown scanner fault. Is
> there any method to get rid of them or ignore? I'm going to make all
> the volumes with noise as a separated task, is it feasible?
> Thanks!
>
> Cheers,
> Emma
>
>
> ---------------------------------------------------------------------------
> Stephen M. Smith, Professor of Biomedical Engineering
> Associate Director, Oxford University FMRIB Centre
>
> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> +44 (0) 1865 222726 (fax 222717)
> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> ---------------------------------------------------------------------------
>
>
>
> ---------------------------------------------------------------------------
> Stephen M. Smith, Professor of Biomedical Engineering
> Associate Director, Oxford University FMRIB Centre
>
> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> +44 (0) 1865 222726 (fax 222717)
> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> ---------------------------------------------------------------------------
>
> <t.zip>
---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------
|