Hello Andreas, Christian, and MELODIC users,
To extract subject-related strength marker we now focussed on the
melodic_mix file that contains the timecurves for all components and
subjects (as a time*subject*voxels matrix). We understood from earlier mails
that these represent timecurves BEFORE scaling by the subject mode.
My question is the values of the melodic_mix files that range between -15 and
15 relative signal amplitude changes (in PERCENT relative to the mean (before
demeaning...)?) reflect the BOLD signal fluctuations for each individual and
component?
Have ALL preprocessing steps defined in group MELODIC run over the original
data before this time courses are extracted (such as high pass filter,
demeaning and variance normalisation?)
Thanks for your help,
best wishes,
Philipp
|