Dear Mark,
I stumbled upon this exchange between you and Jiyang on the list.
While I'm reassured to read that it is possible to perform comparison of more than two groups using FIRST, I am not sure to understand the whole process.
I see how I can compare more than two groups using randomise. What puzzles me is how to manage the different groups in the design matrix that I feed to first_utils --vertexAnalysis.
The matrix example on the FIRST page shows us a design matrix for a simple two groups comparison as a column of -1 x Number of controls and 1 x Number of patients.
What should I enter then, to take in account a third group ? Is a column of 1 x Number of subjects in group 1, 2 x Number of subjects in group 2 and 3 x Number of subjects in group 3 good ?
Or do I need something like
1 0 0
1 0 0
1 0 0
0 1 0
0 1 0
0 1 0
0 0 1
0 0 1
0 0 1
?
Thank in advance for your answer,
best
Alain
>Dear Jiyang,
>Yes, with FSL5 you can set up any design using the Glm tool and test it using randomise.
>Previously it was difficult to do more complicated tests with FIRST, but now that it works via randomise it is just the same >as doing three group comparisons with VBM or FMRI data.
>All the best,
> Mark
On 29 Jan 2013, at 09:07, Jiyang Jiang <[log in to unmask]> wrote:
> Dear FSLers,
>
> I know it is definitely possible to perform 2-group shape analysis using the well developed FIRST tool box. But I was just wondering if the tool box is able to perform multiple group shape analysis, say three groups (disease, control, and at-risk), to test if differences exist on, for example, hippocampal vertices of subjects of these three groups?
>
> Thanks and regards,
> Jiyang
|