You're right that, at least in one situation, theory says it should make little effect. That situation is when you go to do a simple random effects group analysis. It will have an effect if you use the first level estimates of the variance for anything. This includes looking at individual subject significance maps, fixed effects group analysis, or mixed effects group analysis. Also, the new stream separately registers the anatomical and all other time points in the run to the middle time point. The old stream registered the first time point of the first run and motion corrected all runs to that time point.
More information is here: http://nmr.mgh.harvard.edu/~greve/fsfast.intro.ppt
doug
Timothy Vickery wrote:
Hi all,
I noticed in the new version of Freesurfer, you now offer the option to resample raw functional data to a standard (MNI or surface) space at the preprocessing stage (preproc-sess) as opposed to after statistical maps have been computed, but I couldn't find much documentation relevant to this change. I'm just curious -- should this method be preferred for a standard functional contrast GLM analysis? Or is it preferable to follow the old fsfast routine, analyzing in each subject's native functional space and then normalize/concatenate the outputs? It's not completely clear to me that it should make any difference whatsoever in the final product (i.e., group contrast maps), except perhaps with respect to the effects of smoothing...However, I'd be interested to hear the experts' perspectives on this.
Thanks, Tim
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer