Motion Regressors

Hello AFNI Experts,

I have done some initial preprocessing of resting state data in AFNI. Now I plan to do the rest of the denosing and 1st levels for the connectivity analysis in CONN.

I plan to use the motion regressors and censoring files that AFNI produced from my afni proc py script as first level covariates. I will importing the smoothed functional data (not the errts that would have motion regressed out already) and using conn to do the additional denoising steps.

I want to make sure I have I identified the appropriate 1D files, which I think would be:

a.) censor_combined 1D file (0s and 1s, indicating which images should be censored)
b.) motion_deriv 1D file
c.) motion_demean 1D

Are there any other files I should be importing from AFNI in order to properly regress out motion?

Thank you,


These are the usual suspects to be rounded up. Be sure to note that ‘1’ in the censor file means ‘keep this time point’, and ‘0’ means ‘ignore this time point’.

Some people think the file motion_SUBJID_enorm.1D is also useful – it measures the size of the motion. However, I don’t think it’s particularly relevant for regression purposes.

Finally, some pedantic people object to the terms “regressing out motion” or “correcting for motion”, since the effects of the processing don’t actually turn the data you have into the data you WISH you had (e.g., without head movements). Personally, I think this kind of objection is pointless nit-picking, but you might run into it when you present or publish your exciting results.