Hi! I've noticed that, at least in literature using the paradigm I'm using, folks tend to use an 8-mm Gaussian kernel when preprocessing w/ SPM & a 4-mm kernel w/ BrainVoyager or AFNI (only one paper used AFNI).
I previously preprocessed my data using a 4-mm kernel but tried again w/ a 6-mm kernel in light of most papers' using bigger. I found that the maps looked tidier & ROI-based results, more robust (e.g., greater correlations w/ behavioral measures in predicted directions), w/ the 4-mm kernel.
As AFNI defaults to a 4-mm kernel using uber_subject.py, I'm now trying to understand why. Might smaller be better for AFNI & bigger for SPM? How come? Finally, what would you recommend should guide my final kernel selection? Thanks for any thoughts! -Robie
Edited 1 time(s). Last edit at 06/29/2015 06:57PM by neurobie.
