Dear FSL experts,
I am currently analysing resting-state fMRI data from a large data set.
The study has a longitudinal design to invesitgate treatment effects in
patients, but also involves a control group. Three longitudinal scans
were acquired in the patient group, one before treatment and 2 after
treatment at different timepoints. The healthy control group only has
one timepoint.
I am using Melodic and dual regression for data analysis as implemented
in FSL. I was wondering if it is better to temporally concatenate over
all data, all scans and all groups, first, and then define relevant
contrasts in randomize during the dual regression. Or is it better to
run the independent component analyses severately for each and every
group comparison, for example one ICA incluing patients at baseline and
healthy controls, another ICA including patients before treatment and
patients shortly after initiation of treatment. Are both approaches
feasible for data analysis, or does the dual regression require that all
participants from the ICA are included ther as well?
Many thanks for any help you are able to offer.
With best regards,
Patrick Pflanz
Departmental Staff Psychiatry
Department of Psychiatry
University of Oxford
[log in to unmask]
|