 |
 |
Hello all. I am interested in using the voxel dependent EVs option. I would like to have a way of quantifying the effect that including the voxel-wise covariates has on the modeling. I was thinking that having R2 maps for the two models (with and without voxelwise covariates) could be calculated and then the change in R2 would be calculated and tested for voxels having a significant delta R2. Does this seem reasonable? If so I would like to confirm that I would calculate it as: R2 = 1 - (Residual Variance)/(Total Variance) fslmaths filtered_func -Tstd -sqr TotalVariance fslmaths res4d -Tstd -sqr ResidualVariance fslmaths ResidualVariance -div TotalVariance -sub 1 -abs Rsquared I repeat this for both models, voxel-wise subtract the two Rsquared maps and calculate (somehow) an F-test on this map. Thank you for any feedback, Jason -- Jason Steffener, Ph.D. Department of Neurology Columbia University http://www.cogneurosci.org/steffener.html