Hi Darren,
Yes - this all makes sense I think. My only query is about the detail of
the timing of your EVs - it sounds like you set a "1" for the rest volumes
only - that then controls the optimal timing fitting to be for that volume
rather than either of the question or answering - I assume that that is
what you intended.
WRT F-tests - yes, this is a little different from what you are used to.
The F ratio here is the ratio of [ the explained variance in the contrasts
selected to enter the F-test ] to [ the residuals in the full model fit ].
Thus it asks - is the power explained by the selected contrasts
significant?
Thanks, Steve.
On Fri, 16 May 2003, Darren Schreiber wrote:
> In my questions experiment, I have seven different explanatory
> variables (categories of questions). For each of the twenty eight
> questions asked during a session, there are three volumes collected
> (question being asked, answering, and rest (TR=4)). I set up the
> explanatory variables in FSL using the rest volumes. Now, I am trying
> to make sure I understand what I will be getting using different types
> of contrasts. In particular, I am trying to understand the
> implications of demeaning in FSL.
>
> One contrast I could look at is:
>
> A C L M R T U
> A 1 0 0 0 0 0 0
>
> From my understanding, this gives me the mean activation under the A
> condition, having already subtracted out the mean level of activation
> during the entire experiment (ACLMRTU x qar)
>
>
> Another contrast would be:
>
> A C L M R T U
> A 6 -1 -1 -1 -1 -1 -1
>
> My best guess of this contrast is that it is going to show me the
> activations for A that are distinct from activations that are happening
> in CLMRTU.
>
> So, I think that the first contrast would show activations like the
> finger press or the auditory activations. And, I guess I would expect
> that all seven conditions would look pretty similar since I would
> expect the obvious motor or sensory activations to obscure the more
> subtle cognitive differences. And, I would assume that the second
> contrast would take out the common motor and sensory activations and
> highlight the cognitive differences particular to condition A. Is that
> correct?
>
> Further more, how would I use the F test in this case. I am familiar
> with the meaning of F tests when I use them to evaluate whether the
> differences between to subject populations are statistically
> significant. But, I am not clear on the meaning of the maps generated
> by using an F test in FSL.
>
> Thanks in advance,
>
> Darren
>
Stephen M. Smith MA DPhil CEng MIEE
Associate Director, FMRIB and Analysis Research Coordinator
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
|