Hi FSLers,
This is a follow-up on the post below. I am trying to set up a design matrix
to feed randomise. I have carefully read FEAT + randomise documentations and
searched through the archives, but I am still confused.
I have 2 groups (control and experimental, 10 subjects each) and each
subject is scanned in two sessions (pre and post). We expect the
experimental group to show increased FA in the post session compared to pre,
whereas the control group should show less or no increment. So, this is the
main question I would like to answer with my design/contrast(s).
Suppose that I have 3 subjects per group. This is the design matrix that I
finally made:
EV1: group
EV2: session in group 1
EV3: session in group 2
EVs 4-9: subject means
EV1 EV2 EV3 EV4 EV5 EV6 EV7 EV8 EV9
Subj 1, control, pre: -1 -1 0 1 0 0 0 0 0
Subj 2, control, pre: -1 -1 0 0 1 0 0 0 0
Subj 3, control, pre: -1 -1 0 0 0 1 0 0 0
Subj 1, control, post: -1 1 0 0 0 0 1 0 0
Subj 2, control, post: -1 1 0 0 0 0 0 1 0
Subj 3, control, post: -1 1 0 0 0 0 0 0 1
Subj 4, experim, pre: 1 0 -1 1 0 0 0 0 0
Subj 5, experim, pre: 1 0 -1 0 1 0 0 0 0
Subj 6, experim, pre: 1 0 -1 0 0 1 0 0 0
Subj 4, experim, post: 1 0 1 0 0 0 1 0 0
Subj 5, experim, post: 1 0 1 0 0 0 0 1 0
Subj 6, experim, post: 1 0 1 0 0 0 0 0 1
This is made intuitively and I don't know if it makes any sense... If yes,
how could I set up the contrast? Any suggestions, examples for dummies, etc
will be very welcome.
Thanks in advance,
Miguel
- Hide quoted text -
On Mon, Apr 26, 2010 at 9:10 AM, Stephen Smith <[log in to unmask]> wrote:
> Hi,
>
> On 25 Apr 2010, at 22:59, Miguel Burgaleta wrote:
>
> Hello FSLers,
>
> I am working on a pre-post design with a control group and an experimental
> group. The main question we want to answer is whether the experimental group
> shows a higher increment in FA (if any) than the control group. For this I
> have TBSS in mind, but have a couple of general questions that want to share
> with you guys.
>
> 1. Regarding the registration stage... We have 2 acquisitions (pre and
> post) per subject. My idea is to register FA images to a FA template in the
> pre acquisition and, once we acquire the post images, i) to perform an
> intra-subject rigid body registration and then ii) apply the transformation
> matrix + warping obtained in the pre acquisition. Does this sound okay to
> you?
>
>
> Yes it does - hopefully you should be able to amend the tbss scripts easily
> enough to make this not too painful. This kind of longitudinal tweaking is
> on our TODO list....
>
> You could even go further and do something like average the two scans in
> standard space to get the same skeleton-projection for both - but that will
> require more complex fiddling with the scripts, and we're all a little busy
> with conferences right now to offer to help with the details I'm afraid!
>
> 2. Regarding the stats... This is my very first time with this kind of
> design and I am not sure of how to implement this. I have searched through
> the FEAT website looking for a similar design without success. Is there any
> way to integrate between-groups and between-sessions comparisons in a single
> model? And, would it be possible to use randomise in this case?
>
>
> This is simply a two-group paired t-test - you can combine two standard
> paired-t-test designs together (see FEAT and randomise manuals) to get this.
>
> Cheers.
>
>
>
> Any advice will be highly appreciated. Thanks in advance,
>
> Miguel
>
>
>
> ---------------------------------------------------------------------------
> 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
> ---------------------------------------------------------------------------
>
