SPM Estimation fails after freesurfer.BBRegister

When I run the first level estimation on BBRegistered (freesurfer) files it tells me there are no significant voxels. If I use the realigned and unwarped files the estimation succeeds. The only step between realignment and estimation is the boundary based registration. I superimposed the realigned files with the bbregistered and they pretty much overlap. The registration is also very good. I’m running the analysis in nipype.

Does anybody know why the estimation fails? I’m pretty stumped on this one.

I’ve found the problem. I have 7T data with a reduced field of view which means there are lots of zero voxels that make inference difficult. I solved this issue by first cropping the functional runs after BBRegister and also use a mask to limit inference to voxels only inside this field of view.

1 Like