Hi - yes, this should work - it's just a case of persuading FEAT that
all is well. We've just made some minor amendments to the scripts to
allow empty EVs to run through ok and even contrasts involving empty
EVs (effectively assumes that the relevant PE is zero even though
this can't be estimated properly). This will be released in the next
patch release, probably next week.
Cheers, Steve.
On 28 Sep 2007, at 19:30, Eric Claus wrote:
> FSLers,
> I am trying to run first level analyses for each run of a 3-run
> fMRI study.
> Within each run, there are several trial types, but because the
> classification of a trial depends on the participant's response,
> there are
> some cases where 1 or more trial types do not exist in a given
> run. For
> each trial type, I have a separate EV file and for the cases in
> which no
> trials exist, the empty EV option in FSL is chosen.
>
> 2 questions:
>
> 1. When trying to run the subject's model, I get a rank deficiency
> message
> which seems to be expected given the empty EVs. If all the other
> EVs in the
> model look correct (i.e. are not linear combinations of one
> another), are
> the results of the analysis valid?
>
> 2. If I have 4 EVs: EV1 EV2 EV3 EV4 and a given subject has an
> empty EV for
> EV2, how does this affect any contrast with EV2? For example, if a
> contrast
> of 1 -1 0 0 is used, is the resulting statistical map equivalent to
> the
> mean/fit for EV1 or should the stat map be zero, given that the two
> conditions are not really being compared? Further, if the
> participant has
> valid trials for runs 2 and 3 in both EV1 and EV2, when I do the fixed
> effects analysis for this subject, is the run1 contrast included in
> the mean?
>
> Thanks in advance,
> Eric
------------------------------------------------------------------------
---
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
------------------------------------------------------------------------
---
|