We would like to perform functional connectivity on datasets acquired while subjects performed a task. However, we would like to perform the functional connectivity with all task-related effects removed.

I'm following the instructions for Simple Correlation Analysis found here: [afni.nimh.nih.gov]

On Step One, I have a question on which -select option I should use in 3dSynthesize. Would the -select allstim option be the correct option, so that I can use it to remove all task and head motion effects? I assume -select allfunc would not be enough, as it would not include the head motion effects.

As an alternative to Step One, would it be okay to use the -errts output from the original 3dDeconvolve command, and continue following the instructions?



I have one more question, more to confirm I'm understanding the processing correctly.

Once I follow Step One (or use the -errts output?), task has been removed from the time series. I no longer need to worry about it, and do not need to include any task regressors once I get to the actual correlation analysis performed in Step Four. Am I following correctly?

shawn