Hi Matthew,
Thanks for your prompt reply.
Could you perhaps elaborate a bit on your suggestion?
So Running Randomise with the --glm_output flag will produce pe/cope images. Do these index group-level or participant-level estimates?
As mentioned in my initial message, we would like to have them both:
1- PARTICIPANT-LEVEL index (e.g., predicted value or residuals) of the dimensional relationships we documented with VBM & Dual Regression, which we then would feed into another set of analyses.
2- GROUP-LEVEL index of the dimensional relationships we documented with VBM & Dual Regression, in order to better outline/characterise the direction and strength of the associations we find (so more for informative purposes)
3- The pe/cope images you mention are voxelwise data, how can we convert/reduce them to numeric .txt file? And should we only examine pe/cope estimates within the effect sites that emerged in the group analyses.
Hope you can help me out.
Thanks again!
Cheers,
Maurico
Dear Mauro,
If you run randomise with the --glm_output flag ( the number of permutations can be set to 1 for this use case ) then it will output pe/cope images.
Hope this helps,
Kind Regards,
Matthew
--------------------------------
Dr Matthew Webster
FMRIB Centre
John Radcliffe Hospital
University of Oxford
On 12 Nov 2019, at 20:46, Mauricio Delgado <[log in to unmask]> wrote:
Dear FSL experts,
We have run VBM and Dual Regression (Randomise for the stats) on our T1-weighted and RS-fMRI data, respectively, to probe symptoms severity effects on brain morphology and network function.
We have found some interesting relationships between the symptom domains and the VBM/Dual Reg output. We now would like to examine whether the associations we find may have predictive value for changes in several biological indices (all quantitative).
We are, however, not sure how to proceed from here. Specifically, is it possible to somehow extract some index/coefficient (e.g., predicted value or residuals as in regression) for each individual from the VBM/Dual Reg analyses that reflects the brain-symptoms relationships, which then can be used to run post-hoc analyses.
We could obviously extract the mean Grey Matter (VBM) and Z-value (Dual Reg) from the effect sites for each participant, but this does not reflect the actual relationship between symptom severity and brain morphology/connectivity. We would like to have a PARTICIPANT-LEVEL index of the relationships we documented, which we then would feed into another set of analyses.
Also, how can one attain group-level Beta values (and possibly even residuals) corresponding to specific regressors (EV's) and contrasts when using Randomise?
Any assistance would be more than welcome.
Cheers,
Mauro
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|