Compcor in fMRIPrep

1

Dear Experts,

I used fMRIPrep to preprocess my task-based fMRI data. I would like to exclude regressors related to CSF, but keep those reflecting white matter signals.

In the CompCor outputs, I selected covariates starting with ‘c_comp’. Does this mean these components were derived from CSF only, and that the white matter components (w_comp) were not included?

Also, should I include other CompCor components such as t_comp or a_comp in the model?

Thank you very much for your guidance.

Best regards,
Shirley

2

Hi @ShirleyZ,

Yes

No. a_compcor is from the combined WM + CSF mask, and t_compcor is temporal CompCor, which is a different (and not as commonly used) method not typically used in conjunction with anatomical compcor.

Best,
Steven

3

@Steven
For CSF regression without removing the white matter (WM) signal, which approach do you think is better — using the standard CSF regressor or the CompCor (c_comp_cor) regressor?

Thank you very much.

4

I’d personally use the compcor components.

2 Likes