Hi,
I have been working on an seed based resting state analysis and I have a few questions regarding the best way to approach it.
So far, I have:
1) run a preprocessing only analysis on the bold data in feat
- some of my sources recommend doing smoothing and temporal filtering here, but I've seen a post from MJ in the archives saying that only motion correction should be done at this stage. So should it just be motion correction? And then add in the filtering with a subsequent feat?
2) segment CSF and WM regressors from brain extracted T1 structural with FAST
3) run featregapply to get filtered_func_data into standard space
4) run flirt to get output of FAST into standard space
5) fslmeants to extract timeseries from FAST output
- I have seen some people seem to be moving everything into functional space instead of standard space for this. Which is recommended and how does it affect the downstream processing?
6) run post-stats feat with reg_standard/filtered_func_data with glm for motion outliers, CSF, WM, and motion correction regressors.
- If I have only run motion correction in the preprocessing feat, do I need to do a full analysis here, adding the temporal and spatial filtering back in?
I also have several runs of the resting scan for the same patient. What is the best way to average these? Can I run a fixed effect analysis on just the prestats? Or should I run on the post-stats output instead? What image do I need to work with after the runs have been averaged?
Now I know I need to plug things into a dual regression. I think I'm supposed to be using stats/res4d.nii.gz from the post-stats analysis from each subject. Is that correct?
It seems that the command dual_regression is geared toward ICA output. Can I use it for seed-based? It was suggested to me by a colleague that I could sphere my 3D ROI mask to make it 4D and plug this in. Would that work?
I have to say I'm quite confused about to set up this step, so any guidance would be helpful. We're using an ROI from a previous analysis as a seed to compare between a control group and a patient group.
I know this is a lot of questions, so thank you very much in advance!
Best wishes,
Erin Drazich
|