Hi Steve,
the design matrix was an example for a paired t-test of 5 people (to
make the matrix shorter - i changed the matrix, but stick to the
example with 5 and hope now it makes more sense and that is indeed the
correct design.matrix).
And yes - the answer to the previous FLAME contrast-of-contrast
question did clarify the situation a bit. However, maybe let me try to
rephrase:
I assume that negative numbers in the output from siena
(A_to_B_flow.nii.gz) mean that there has been atrophy in this region,
while positive mean that growth occured.
>siena_flow2std A B< just transforms these into standard space (and
keeps the meaning of negative numbers being atrophy and positive being
growth).
Now if I test for differences between the drug and the placebo
condition with this design matrix and contrasts.
Input Group EV1 EV2 EV3 EV4 EV5 EV6
s1_prepost_drug 1 1 1 0 0 0 0
s2_prepost_drug 1 1 0 1 0 0 0
s3_prepost_drug 1 1 0 0 1 0 0
s4_prepost_drug 1 1 0 0 0 1 0
s5_prepost_drug 1 1 0 0 0 0 1
s1_prepost_plac 1 -1 1 0 0 0 0
s2_prepost_plac 1 -1 0 1 0 0 0
s3_prepost_plac 1 -1 0 0 1 0 0
s4_prepost_plac 1 -1 0 0 0 1 0
s5_prepost_plac 1 -1 0 0 0 0 1
Contrasts
EV1 EV2 EV3 EV4 EV5 EV6
Condition A>B 1 0 0 0 0 0
Condition B>A -1 0 0 0 0 0
So my understanding is that Contrast_1 (Condition A>B) is asking for
where has more growth (or less atrophy) happened during the drug
condition compared to placebo?
And Contrast_2 (Condition B>A) is asking the same thing for the
placebo condition compared to the drug condition?
Is my interpretation right? Thanks a lot, Michael
On 17-Feb-09, at 12:50 AM, Steve Smith wrote:
> Hi,
>
> It's not clear how your design relates to your "20" drug/placebo
> timepoints - however maybe this is answered by my previous email,
> about FMRI contrasts-of-contrasts?
>
> Cheers.
>
>
> On 17 Feb 2009, at 00:27, Michael Scheel wrote:
>
>> Dear fsl experts,
>>
>> I have a question regarding the output of siena. We scanned 20
>> subjects at four different timepoints - 1) PreDrug 2) PostDrug
>> 3)PrePlacebo 4) PostPlacebo.
>> I used siena to estimate atrophy/growth in both conditions (Placebo
>> and Drug) individually
>>
>>> siena subject_predrug.nii.gz subject_postdrug.nii.gz
>>> siena subject_preplacebo.nii.gz subject_postplacebo.nii.gz
>>
>> As suggested on the website I then did >run siena_flow2std A B< and
>> used fslmerge to merge all the A_to_B_flow_to_std.nii.gz into one 4D
>> file (all_flow_to_std.nii.gz) so that first all the 20 were flow
>> images between Pre and Post for the DrugCondition and the last 20
>> were the changes under the Placebo Condition. I then set up matrix
>> and contrasts with the Glm_gui wizard to set up a paired ttest and
>> used randomise with the threshold-free-enhancement-option.
>>
>> As the whole procedure was rather to control that there wasn't any
>> particular atrophy so I didn't expect any significant results.
>> However I do get signifant voxels for the Contrast 'Condition B >
>> Condition A' Contrast and was wondering how to interpret these
>> findings, would it mean that the atrophy is bigger in the Placebo
>> Condition?
>>
>> Below my design matrix and contrast (example for 5 subjects)
>>
>> Input Group EV1 EV2 EV3 EV4 EV5 EV6
>> 1 1 1 1 0 0 0 0
>> 2 1 1 0 1 0 0 0
>> 3 1 1 0 0 1 0 0
>> 4 1 1 0 0 0 1 0
>> 5 1 1 0 0 0 0 1
>> 6 1 -1 1 0 0 0 0
>> 7 1 -1 0 1 0 0 0
>> 8 1 -1 0 0 1 0 0
>> 9 1 -1 0 0 0 1 0
>> 10 1 -1 0 0 0 0 1
>>
>> Contrasts
>> EV1 EV2 EV3 EV4 EV5 EV6
>> Condition A>B 1 0 0 0 0 0
>> Condition B>A -1 0 0 0 0 0
>>
>>
>> Thanks, Michael
>>
>
>
> ---------------------------------------------------------------------------
> 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
> ---------------------------------------------------------------------------
|