Hi. I’d like to use nipype’s CompCor interface but I’m not sure how to use it. Basically I would like to use the interface to clean up my resting state data as discussed in Behazadi et al. 2007. That means regressing out the WM and CSF and possible other confounds such as movement. Is this possible using nipype’s implementation of CompCor? How can I input the WM and CSF maps as well as movement regressors and run the regression?
Thanks!
I think I figured out how to run the ACompCor interface. Basically I passed the white matter and csf masks to the “mask_files” input, is that correct? I used the anatomical segmented white matter and csf masks as “mask_files” and for the “realigned_file” I used the realigned EPIs. This however raised an exception because the EPIs and anatomical masks don’t have the same dimensions. I then put the anatomical segmented masks into functional space and it worked.
So now I have a “components_file”. My question is how can I regress out these components from the signal of interest in nipype?
If you set up the appropriate model, yes, you can regress those components out. (see http://nipype.readthedocs.io/en/latest/interfaces/generated/nipype.algorithms.modelgen.html#specifymodel and examples making use of it, e.g. http://nipype.readthedocs.io/en/latest/interfaces/generated/workflows.fmri.html).
How to generate confounds timeseries .json and .tsv using fmriprep compcor? - Neuro Questions - Neurostars I am trying to do something similar it would be great if I could get some guidance.