Hi,
The GLM, when used with an appropriate contrast (just on the regressor
associated with the data to be correlated) will calculate the p-value
associated with testing if the slope is non-zero. The command line call
here looks fine as long as your design matrix is correct (see examples in
the GLM page of the FSL wiki).
All the best,
Mark
On 07/01/2015 17:25, "SUBSCRIBE FSL Anonymous" <[log in to unmask]>
wrote:
>Dear FSL experts,
>
>I'm a little confused with the outputs of fsl_glm, between correlations
>and regression, and associated p-values
>
>I want to correlate GM maps and behavioural scores and I want to obtain
>the p-values of the correlations. Is the following line correct?
>
>fsl_glm -i input.nii.gz -o output.nii.gz -d matrix.mat -c contrast.con
>--out_p=p.nii.gz --demean --des_norm --dat_norm
>
>And, if I aim to obtain the p-values of the slopes, which would be the
>correct command line? Or the line I wrote above outputs the p-values of
>the slopes?
>
>thanks a lot for your support.
>
>Marco
|