Coregistering a partial functional image to a full structural image

I have a full structural image and a functional image with the bottom half of the cerebellum cut off (i.e., the func image covers 95% of the brain).

When registering func to struct (or to standard), is there anything I should keep in mind? So far I’ve tried traditional FSL FLIRT and FNIRT, as well as BBR. Both outputs look okay, but not fantastic.

Most functional EPI datasets do not include at least part of the cerebellum. Things to consider when aligning EPI to anatomical data (or vice versa) include tissue contrast, voxel resolution and smoothness, coverage, nonuniformities, skull, anatomical type (T1/T2), distribution of EPI voxel values, how far apart to start,… These will influence the parameters used for computing the alignment including cost functions. AFNI includes align_epi_anat.py for doing this. Within a typical FMRI process, you would normally use afni_proc.py that would call all the other necessary programs. See the PDF below for more info about all these considerations.

Thanks! This is a really thorough explanation of registration in AFNI. Is AFNI the best tool for coregistering partial images? I’m currently using FSL; should I make the switch?

I’m not as familiar with the other tools, so I will leave that up to you to try. AFNI includes “-partial_…” options for dealing with partial datasets. If the data is “too partial” though, then that can be difficult.