Print

Print
Hi

On 9 Dec 2015, at 15:47, Tae-Ho Lee <[log in to unmask]> wrote:

Dear experts,

I've been working on resting state data using dual regression, and have several questions with results. I've tried to find (and post) threads in both FSL forum and google about those but fail to get clear answers about it. I would appreciate any insights for my questions.

So far, what I did was that,
1) Ran concat ICA and did template matching procedure.
2) Successfully identified RSN components (DMN, FPN, ECN ... )
3) Fed "mean_IC.nii.gz" into dual regression process
4) Got individually estimated time-series files (stage1) and zmap (stage2)

At this point, my questions are
1) When I feed "mean_IC.nii.gz" into dual regression, I wondered it is okay to threshold the mean_IC.nii.gz with a certain z (or p) value. It seems that dual regression is using just whole component activities (instead of restricted activities at a certain threshold), and therefore individual time-course was estimated from broader boundaries of component. So I thought that if I constrained the group component map for dual regression (with a higher threshold), then the final estimated time-course would be more specific within the network component.  

that's largely empirical I think - within reason it could make sense, though it's not very obvious what the practical pros/cons would be.


2) Instead of conducting group-level analysis in dual regression to see group differences (stage 3), I would like to estimate the connectivity strength, at the individual level, between one of RSN components and the other structural region of interest. For example, I want to see how much individual DMN component (stage 2) is strongly connected to the their amygdala activity.  This is basically I want to use the connectivity strength as individual index or difference values. In this case, I wondered whether it is okay to extract mean signals from individual component z-map (stage 2) with several structural ROI masks, and consider it as individual functional connectivity strength between RSNs and ROIs.

HI - you can extract regional summary measures from a given dual-reg spatial map output - but this *only* tells you about "connectivity to" this RSN - not all others - so it is not a general measure of connectivity from a given region to the rest of the brain in the context of the full set of resting-state networks.

Cheers





A few days ago, I also posted another question regarding group-ICA, but I've not yet heard anything. I would appreciate if you have a chance to look at my previous question as well. Here's the link.
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1512&L=FSL&F=&S=&X=5D55B9561D691C1EDD&Y=taehol%40usc.edu&P=70713


---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Head of Analysis,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------

Stop the cultural destruction of Tibet