On 11/12/2021 3:29 AM, Alice Giubergia wrote:
External Email - Use Caution
Thanks for your replies Doug! I copied the previous messages below, since I did not menage to retrieve the original email.
- Thanks.
- Regarding question number 2, yes that was what I meant.
- I ran the task analysis and extracted time series from unwhitened
residuals by mri_segstats, upon a specific segmentation. I was thinking now of a correlation matrix by Pearson's coeff. Any further suggestions?
No, Pearson is fine
Also, I've got a pheraps silly question:
- When running a task analysis, I could specify -taskreg and -nuisreg
separetly. My assumption is that only -nuisreg are regressed out from the data and that the residuals saved got task-related bold signal. Is it correct?
No, this is done simultaneously. There are a few difference. For nuisreg, you will have separate regressor sets for each run (ie, different regression coefficients for each nuis for each run). For taskreg, you get one for each taskreg for each run. You also cannot compute a contrast for the nuisreg.
- If so, could I use task regressors as -nuisreg to get a "pseudo"
resting state?
You can do the task-only analysis and save the residual as I described before. This will remove the task signal.
Hoping all question are clear, Thanks for your patience, Alice
Dear FreeSurfer experts, I'm interested in performing a functional analysis on task-based data. I was taking as an example the "FsFastFunctionalConnectivityWalkthrough" tutorial but, if I got it correctly, is specific for resting state data. So I've got some doubts and questions: 1) Given the steps described in the above mentioned tutorial, the file "pcc.mgz" generated in the session folder gives me the correlation strength between the time series of each voxel and the seed (L_Posteriorcingulate in the tutorial)?Yes, correct
2) If I want to make a functional analysis on task data, similar to the one described in that tutorial, could I just add the -event-related option instead of -notask option? (mkanalysis-sess -analysis fc.lpccseed.surf.lh -surface fsaverage lh -fwhm 5* -notask* -taskreg L_Posteriorcingulate.dat 1 -nuisreg vcsf.dat 5 -nuisreg wm.dat 5 -mcextreg -polyfit 5 -nskip 4 -fsd bold -TR <TR>) Do you mean you want to do a simultaneous analysis of the task and the connectivity? I just tried what you suggested and it looked like it worked. 3) Otherwise, does it make sense to you If I run a simple task analysis to regress nuisances out (without seed region), extract time series and get the pair-wise correlation with Pearson coefficient?This is possible, and makes sense. It is a little difficult to do, but not terrible. If you want to go this route, you would need to first analyze your task data as task without FC, specifying -svres-unwhitened to save the residuals. Let me know when you get to that point and I can push you further.
Hoping my questions are clear, Thanks a lot for your patience and help, Alice
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer