Hi, all. I've been cross-checking some my data analysis from SPM with
FSL, and it's been very useful for finding some (incorrect) settings
I'd forgotten I'd made.
However, there's one thing that I can't figure out how to do in FSL
that's easy in SPM: adding a non-standard number of nuisance
regressors per subject. Using the artifact rejection toolbox in SPM
(http://www.fil.ion.ucl.ac.uk/spm/ext/#ArtRepair), you can clean up
your data by adding entire volumes as nuisance regressors. So if you
have a few signal intensity spikes that you'd like to keep out of your
residual, you can add an individual nuisance regressor for each
outlier volume in a run.
This works in SPM because it's simple to write scripts that build
contrasts using the regressor names, rather than the regressor
indices. The fact that different subjects may have very different
numbers of regressors doesn't stop you from doing batch analysis.
I can't figure out how to do a similar thing in FSL. To batch my
analyses, I've created a template .fsf file, then (as suggested by
this mailing list), used simple scripts to search and replace
subjectIDs and directory names to create individual feat setup files.
This works fantastically when every subject has the same number of
EVs, but how do I make it work when subjects have different numbers of
EVs?
Thanks!
Todd
|