Hi All,
We are trying to speed things up in our functional analysis pipeline, and I wanted to consult with the experts here about the following suggestion...
At the moment, we typically collect multiple runs per subject (say, 10-20). We usually then run pre-processing on each run, and then do first-order analysis on all available runs. This step is obviously a bottleneck because all runs need to be loaded to memory and the analysis takes a long time.
As an alternative, we were thinking about doing 1st order analysis on each individual run, and then combine them somehow using 2nd order group analysis.
My silly questions are: 1. Does this workflow make sense? 2. How should we go about and do this? My initial thought was to use mri_glmfit, and to build a 4D volume that will contain the beta's of interest from each individual 1st order analysis.
Thanks,
-- Shay