Print

Print
Hi Colleen,

You'll need to have a design matrix and contrast file for each subject (1st level) and another design matrix and contrast file for the conditions and subjects (2nd level). Both of these can be created with the command "Glm" (in Linux) or "Glm_gui" (Mac).

For the 1st level, use the option "Time series design". The documentation for all the options is in the FEAT manual. You may also want to do the practicals in the FSL Course to get a grip.

For the 2nd level, use the option "Higher-level / non-time series design". This 2nd design includes repeated measures and, depending of the hypothesis, it may not be among the examples in the GLM manual. If you need help with it, please, give details of what the hypothesis is.

All the best,

Anderson


--
Anderson M. Winkler
FMRIB / Analysis Group
[ Blog | Twitter ]


On 2 September 2015 at 15:56, Colleen A. Hanlon, PhD <[log in to unmask]> wrote:
Dear FSL community,

I have 2 fairly basic question about specifying a design matrix in FSL to be used as post-stats on a Melodic Tensor ICA analysis.  The documentation leads me to the FEAT GLM GUI, but Im not sure which of those models apply in this situation.

I have run MELODIC group tensor ICA on a data set of 40 individuals (2 groups) that were performing a basic block design task (~30 sec blocks - with 3 task conditions).    All subjects have exactly the same order.   If I were using SPM, the design matrix for a single run within subject might be something theoretically like:

Condition 1, Subject 1: 000011111000001111100000
Condition 2, Subject 1: 111100000000000000011111
Condition 3, Subject 1: 000000000111110000000000


1) Using Feat should I just specify a design matrix like this for one individual (what we would call *first level* in SPM) and then load that into MELODIC post-stats?    If so , will it automatically apply this design to all participants?

2) Assuming the above is correct, how do I indicate that 20 of the individuals are in group 1 and 20 are in group 2?

Thank you very much for your time.    Its always fun to be learning new things..

Best,
Colleen


(I could also just specify it manually, but Im not sure of the format FSL expects the design matrix to be in)