HI YM Huh,

You can set up a GLM with the demeaned behavioural scores using the GLM gui and save this (for more information: https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#Single-Group_Average_with_Additional_Covariate). Then provide the saved design.mat and design.con from this in your dual regression call. When looking at the dual regression results, make sure to look at the contrast of interest.

Please note that the components that you feed into the dual regression should be estimated using concatenated group ICA.

Hope that helps.

Best wishes,

Janine

-----
Dr Janine Bijsterbosch
Postdoctoral Researcher
FMRIB Centre, University of Oxford
John Radcliffe Hospital
Oxford, United Kingdom
[log in to unmask]ac.uk

On 5 Dec 2016, at 11:05, YM Huh <[log in to unmask]> wrote:

Hi,
I have preprocessed resting state fMRI data, and found a few resting state components using commandline "melodic" as below.
(melodic -i filename -o result_ica -v --nobet --tr=3.5 --report --guireport=./report.html --mmthresh=0.5 --Ostats -a concat)
I want to extract individual temporal information using dual regression, and test correlation with behavioral scores.
How can I do this? I only have a single group, so I can't test for group difference which dual regression basically provides.

Best regards,

YM Huh