Hi,
The Glm GUI in this context creates a design.grp file which you can
feed into randomise. Yes, the grouping is different for randomise than
it is for FLAME, as described in the randomise manual: you would have
a single column in this, with '1' for each block of data that you want
to permute within.
Cheers, Steve.
On 21 Feb 2008, at 07:18, bs Jeong wrote:
> Hi Steve,
>
> Thank you for your comment about triple t-test.
> I need to perform group by time interaction in three group ( having
> 3 time
> points in each group) comparison (FSL-VBM).
>
> To do this, I think all images (3 time points/subject x 3 subject/
> group x
> 3 group) put into one GLM model for triple T-test. And to identify
> each
> group, I think, I have to make 'group.dat' file (which consist of
> nine 1s,
> nine 2s, nine 3s) rather than to define group using group variable
> in GLM
> model window. And then I could results of time1-time2, time2-time3,
> time1-time3 of each group and results of group comparison (group A -
> group
> B, group A - group C, group B - group C). Am I correct? If so, what
> I have
> to see or do for group by time interaction?
>
> thanks,
>
> BumSeok
>
>> Hi,
>>
>> On 17 Feb 2008, at 16:44, bs Jeong wrote:
>>
>>> Dear FSL experts,
>>>
>>> I have a question for performing FSL-VBM (3 groups * 3 time point).
>>> 1. To make template, do I have to select same number of images from
>>> 3 time
>>> point as well as 3 group?
>>
>> You could just use one time point from each subject from each group
>> or
>> use all 3, as long as the numbers are matched.
>>
>>> 2. I have 3 diagnostic group and each subject has 3 images which
>>> were get
>>> different time. My hypothesis is that GM intensity will be decreased
>>> across
>>> the time ( 1 vs 0 vs -1 for linear decline)and the slop of decrease
>>> will be
>>> different among 3 groups (1 vs 1 vs -2: former 2 groups will not be
>>> different but GM intensity of 3rd group will be more decreased.) and
>>> there
>>> will be group by time interaction. But to make design matrix (for
>>> main
>>> effects and group by time interaction) is not easy for me.
>>> If each group has 3 subject, I think design matrix like below. But
>>> I'm not
>>> sure it is correct and don't know the matrix for group by time
>>> interaction.
>>
>> I don't think your example is correct - I think you want the
>> tripled-t-
>> test example in the FEAT manual (see higher-level examples).
>>
>>> 3. And do I have to make group.dat file which consist of nine 1s,
>>> nine 2s,
>>> nine 3s?
>>
>> Yes, see the randomise manual for information about repeated
>> measures.
>>
>>
>> Steve.
>>
>>
>>>
>>>
>>> I'd be very happy if I can hear any comment.
>>>
>>> thanks,
>>>
>>> bsjeong
>>>
>>> sj1 sj2 sj3 t1 t2 t3 time
>>> group
>>> gr1_sj1_t1 1 0 0 1 0 0 1
>>> 1
>>> gr1_sj1_t2 1 0 0 0 1 0 0
>>> 1
>>> gr1_sj1_t3 1 0 0 0 0 1 -1
>>> 1
>>> gr1_sj2_t1 0 1 0 1 0 0 1
>>> 1
>>> gr1_sj2_t2 0 1 0 0 1 0 0
>>> 1
>>> gr1_sj2_t3 0 1 0 0 0 1 -1
>>> 1
>>> gr1_sj3_t1 0 0 1 1 0 0 1
>>> 1
>>> gr1_sj3_t2 0 0 1 0 1 0 0
>>> 1
>>> gr1_sj3_t3 0 0 1 0 0 1 -1
>>> 1
>>> gr2_sj1_t1 1 0 0 1 0 0 1
>>> -2
>>> gr2_sj1_t2 1 0 0 0 1 0 0
>>> -2
>>> gr2_sj1_t3 1 0 0 0 0 1 -1
>>> -2
>>> gr2_sj2_t1 0 1 0 1 0 0 1
>>> -2
>>> gr2_sj2_t2 0 1 0 0 1 0 0
>>> -2
>>> gr2_sj2_t3 0 1 0 0 0 1 -1
>>> -2
>>> gr2_sj3_t1 0 0 1 1 0 0 1
>>> -2
>>> gr2_sj3_t2 0 0 1 0 1 0 0
>>> -2
>>> gr2_sj3_t3 0 0 1 0 0 1 -1
>>> -2
>>> gr3_sj1_t1 1 0 0 1 0 0 1
>>> 1
>>> gr3_sj1_t2 1 0 0 0 1 0 0
>>> 1
>>> gr3_sj1_t3 1 0 0 0 0 1 -1
>>> 1
>>> gr3_sj2_t1 0 1 0 1 0 0 1
>>> 1
>>> gr3_sj2_t2 0 1 0 0 1 0 0
>>> 1
>>> gr3_sj2_t3 0 1 0 0 0 1 -1
>>> 1
>>> gr3_sj3_t1 0 0 1 1 0 0 1
>>> 1
>>> gr3_sj3_t2 0 0 1 0 1 0 0
>>> 1
>>> gr3_sj3_t3 0 0 1 0 0 1 -1
>>> 1
>>>
>>
>>
>> ---------------------------------------------------------------------------
>> 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
>> ---------------------------------------------------------------------------
>>
>
---------------------------------------------------------------------------
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
---------------------------------------------------------------------------
|