Dear Muhammad,
On Wed, Feb 24, 2010 at 4:02 PM, MP <[log in to unmask]> wrote:
> Thanks Donald and Will,
>
> Just to follow up on Will's suggestion: Can I use Imcalc for the EEG
> Time-Frequency images as well?
The images that are generated by 'Convert to images' are in the NIfTI
format which is the same as for any other images processed by SPM. So
the answer is yes.
Also, in the F-contrast [1 -1 0; 0 1 -1] that
> you suggested, how is it testing for the differences between C and A. Should
> I use [1 -1 0; 0 1 -1; -1 0 1] as the F-contrast instead?
>
These contrasts are equivalent because you can get the second row by
adding the third one to the first (and multiplying by -1).
Best,
Vladimir
> On Wed, Feb 24, 2010 at 10:39 AM, Will Penny <[log in to unmask]>
> wrote:
>>
>> Dear Muhammad,
>>
>> What I would do is first create an A-B difference image for each subject.
>> You can do this with imcalc. This then constitutes the data for your 2nd
>> level analysis.
>>
>> Then for your 2nd level design use the 'full factorial' and specify
>> one factors with 3 levels. These will be your 3 groups.
>>
>> To test for the difference effect in group A use [1 0 0]. F for two-sided,
>> t for pos or neg.
>>
>> Similarly for group B use [0 1 0] and [0 0 1] for C.
>>
>> To see where in the brain the difference effect is different between
>> groups use the F-contrast [1 -1 0; 0 1 -1]. This is a two-dimensional
>> contrast where the first dimension (or row) tests for any differences
>> between A and B. Second row for differences between B and C. The F-contrast
>> tests for any linear combinations of the two.
>>
>> You can then use SPM's various plotting functions to see what is driving
>> the effect (it could be eg. just differences between B and C - of course you
>> can then explicitly test this with a t-contrast).
>>
>> Best,
>>
>> Will.
>>
>>
>> MP wrote:
>>>
>>> Hello SPMers,
>>>
>>> I am currently analyzing Time-Frequency EEG data. My design has 3 groups
>>> (X, Y, Z, with 10 subjects in each group) and 2 conditions (A, B), so 2 TFRs
>>> per subject. I want to know how Time-Frequency data differs between
>>> condition B and condition A for each group and also how this difference
>>> (B-A) differs between X, Y and Z groups.
>>>
>>> For this I first tried to use the paired t-test separately for each
>>> group, where I identified each subject as a pair of A and B conditions, so
>>> 10 pairs altogether. After estimating the design matrix I could not come up
>>> with the contrast value that will give me my desired contrast.
>>>
>>> I also tried to use the full factorial design, to also find out the
>>> between group differences. Here I specified 2 factors (Condition with 2
>>> levels and Group with 3 levels). I then also loaded the images in the
>>> corresponding cells. Again, after the estimation of the design matrix, I
>>> could not come up with the contrast value that would produce the desired
>>> contrast.
>>>
>>> Could someone please guide me how to specify these contrasts, and if
>>> there is a better way to set up the design matrix.
>>>
>>> Many thanks
>>>
>>> -Muhammad Parvaz
>>
>> --
>> William D. Penny
>> Wellcome Trust Centre for Neuroimaging
>> University College London
>> 12 Queen Square
>> London WC1N 3BG
>>
>> Tel: 020 7833 7475
>> FAX: 020 7813 1420
>> Email: [log in to unmask]
>> URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
>>
>
>
|