Hi,
The example you sent defines the following:
First event starts at 12.01 seconds and lasts for 3 seconds
(finishing at 15.01)
Second event starts at 30.025 seconds and lasts for 6 seconds
(finishing at 36.025)
Third event start at 48.041 seconds and lasts for 8 seconds
(finishing at 56.041)
If you added another EV which modelled *all* of the other times (i.e.
from 0 to 12.01
then 15.01 to 30.025 and then 36.025 to 48.041 etc.) then it would be
rank
deficient because the sum of the two EVs would be a constant, and the
mean
constant level is removed in FSL and hence cannot be modelled (this is
different
to SPM). The bottom line is that the rest/baseline/fixation condition
*must not*
be modelled in FSL.
Note that rest/baseline will always be the implicit contrasting
condition to any single
EV that you place a contrast on. In your original email you tried to
ask about
both the difference between stimuli and rest and, separately, the
"impact of the
stimuli". The latter question has no meaning in FMRI. You *must
always* compare
between two conditions in FMRI. You cannot just look at one condition
alone,
in the same way that an experiment where there is one experimental
condition
and no rest or baseline condition cannot be anaylzed, since it is only
differences
in BOLD signal which have meaning.
In terms of setting up your analysis, it is fairly easy. Just model
the stimuli and
not the rest/baseline/fixation. I hope the information above makes
the 3 column
format clear.
All the best,
Mark
On 1 Nov 2008, at 09:44, Yael Shani wrote:
> Dear Mark
> I read your mail again and I'm afraid I don't understand why the
> model with the rest included was not supposed to work.
> If it's alright, I want to make sure I defined the events right.
> The stimuli was defined as event 1. since it is an event related
> design I used a Custom 3 file. An example for the rows written is
> below:
> 12.01 3 1
> 30.025 6 1
> 48.041 8 1
>
> which means that the stimuli has started in the 12th second, was
> displayed for 3 seconds, and for the next 15 seconds or so a
> fixation was displayed.
>
> (this is with one event). Did I get it right?
>
> Now, if I defined a Rest event, which is the complement time of the
> above event, how does fsl know that it is a rest event and not be
> able to analyze the contrast?
> I'm asking this question because, after reading your mail, I am not
> sure I defined my event correctly.
>
> Thank you again for your help
> Yael
> On Sat, Nov 1, 2008 at 9:29 AM, Mark Jenkinson <[log in to unmask]>
> wrote:
> Dear Yael,
>
> I wouldn't worry about the error or not.
> It is wrong to model the baseline/rest/fixation condition
> so just don't do it.
>
> As for the website, thanks for letting us know.
> It is dated February 2006 and so, being over two and a
> half years old, is naturally out of date. If you want to
> know how the current version of FSL works, look at the
> online documentation, FSL course and FAQ on our
> website (www.fmrib.ox.ac.uk/fsl).
>
> All the best,
> Mark
>
> On Fri, Oct 31, 2008 at 9:58 PM, Yael Shani <[log in to unmask]>
> wrote:
> Dear Mark
>
> 1. I haven't received any notification of an error, but I will check
> again.
> 2. Regarding hdr. - I saw an explanation about the hdr in the
> website of Afni: http://www.sph.sc.edu/comd/rorden/workshop/fsl/efmri/
> I don't think that I saw any reference in the fsl website.
> Thank you
> Yael
>
> On Fri, Oct 31, 2008 at 6:49 PM, Mark Jenkinson
> <[log in to unmask]> wrote:
> Dear Yael,
>
> You should not model the rest/baseline/fixation condition in FEAT.
> Whatever you did to set up the model with it included should not
> have worked (it should have complained about it being rank deficient)
> so it probably just wasn't set up correctly (with the right durations
> or timings).
>
> As for image format, the *preferred* format for FSL is NIfTI which
> has an extension nii (or nii.gz if it is compressed). This is what
> we say that people should use on our documentation and webpages.
> Why did you think you needed a hdr file? This is the old Analyze
> format which we now discourage people from using. If you have
> found a place where it says you should use this then please let us
> know so that we can update it.
>
> All the best,
> Mark
>
On 31 Oct 2008, at 16:45, Yael Shani wrote:
> Dear fsl-ers
>
> I have two separate questions regarding FEAT analysis for an event
> related design:
> 1. As far as I understand from the tutorial guide of FEAT, for an
> event related design I do not need to define the rest (the Fixation
> display) as an event. Am I right?
>
> I tried to play with the data and just for the curiosity I did
> define the rest as an event. I ran two contrasts: one examine the
> impact of the stimuli (1 0) and the other is the difference between
> the stimuli to the rest condition (1 -1).
> I received different results in the FEAT display for both
> conditions. Why is that?
>
> 2. The second question deals with registration. I transformed the
> data (including the anatomical data) from DICOM to AFNI and
> recieved the files with nii ending and not hdr.
> Can I use the nii file? If not, how can I transform the nii file to
> an hdr file?
>
> Many thanks
>
> Yael
|