Hello,
I've been trying to evaluate the quality of registration of my functional data to 2mm standard space and noticed that the reference/red lines for the standard in the reg summary are leaving out some of the functional images' voxels [see attached images - example slices cropped from the summary registration image; representative of most of my subjects]. I've been running MELODIC with nonlinear registration @ 10 mm warp resolution and passing the anat_brain image into "Main structural image" using Normal search and BBR. Looking through other subjects' registrations, I've noticed that both the functional to high-res and the functional to standard registrations are failing to envelope the functional image, particularly in the frontal cortex. I have a few questions in general about the registration:
1) How should I interpret the reference lines? Is everything that lies outside of those lines excluded from subsequent analyses?
2) I've been playing around with BET's options and I've determined that, depending on the stringency of voxel inclusion, I'll either have some subjects with some residual scalp/skull [more conservative approach; eg., -f .1 -B] or subjects with some bits of the cortex eroded [e.g., -f .3 -B]. In general, would you advise on prioritizing cortex inclusion over scalp removal for MELODIC registration and analysis?
3) I understand that the registrations won't be perfect; are the registrations I've attached on this message comparable to what you/others achieve?
Thanks for your time and for this great software package!
Riley
|