Hi All,

I am looking for a fsl method to correct for multiple comparisons in data where the voxel-wise degrees of freedom differ from voxel to voxel due to presence of EPI drop out.
The standard approach is to restrict analyses to voxels in which all subjects have values. This works in small samples, but in samples of N>200, this **considerably** shrinks the analyzed region, and cannot be fixed just by dropping one or two offenders.
I am unaware of a good approach to dealing with this problem. I can combine subject specific masks to compute the group dof for each voxel easily enough, but I am unsure of where to go from there.

Sincerely,

Joshua