Hi Paul and thanks for your fast reply!
Yes, I'm using randomise to calculate my statistics. What I'm exactly doing is comparing MD maps between two groups in a voxel-based basis. As I've said, some subjects are spatially cut (superior part of the brain).
So, if I have undestood your explanation correctly... Imagine I have 100 subjects, and subjects #23 and #45 are cut.
1- The randomise input is going to be the 4D volume concatenating all MD maps.
2- I have to create two adittional 4D volumes, one per cut subject. The first one for the subject #23, in which all volumes are 0 except the volume number 23, which is going to be a binary mask. Same procedure for the other subject, a 4D containing all 0 but a binary mask at the index #45.
3- Add two 0 columns to the design matrix. Imagine I had 2 columns, now I have 4.
4- Pass the index numbers in the option --vxl 3,4
5- Pass the 4D additional volumes "--vxf subj23_mask.nii.gz,subj45_mask.nii.gz".
Am I wrong?
Thanks again.
Eduard.
PS: By the way, does it take longer to finish than a normal randomise?
|