I would like to extract some FEAT parameters from a region of interest. Specifically, I would like to extract the parameter estimate (cope), its variance (varcope), and the t-value (which can be calculated from the first two).
I can see two ways of going about this:
1) Calculate the average time-series within an ROI. Then, run a GLM with the covariates of interest and extract the beta, var(beta) and t-value. I'd probably use Matlab for this, as I don't know how to do this in FSL.
2) Run the standard FEAT pipeline and extract the cope and varcope for each voxel in the brain. Then, estimate the beta as the average of the cope across voxels. However, I wouldn't know how to calculate the variance of this beta within the ROI, given that the voxels in the ROI are not independent.
Can anyone suggest me the best way of doing this?
Thanks!
|