Hi ASI,

1. What do you mean it is not precise enough? Why is the
correlation value not what you want?

Are you complaining that the result is not stored in a
dataset that you can access from the command line? If
so, just save the overlay (Define Datamode -> SaveAs -> OLay).

2. That is right. Having processed the data with afni_proc.py,
your input to InstaCorr would be the errts dataset, and:

- no blur
- no Automask (you must choose a mask though, probably full_mask)
- no Bandpass, etc.

Note that you need to actually turn on "Misc Opts" and set
Polort to -1, as 2 is the default.

3. If you wish a seed radius, set Misc Opts -> SeedRad. Note
that if you turn on Misc.

But this seed would not be the ROI cluster average, it would be
a spherical average centered at the seed. InstaCorr has no
option to take an ROI for a seed (which would mostly defeat the
purpose of using InstaCorr (i.e. having a known seed/ROI)).

Note that 3dGroupInCorr does have such an option.

I suggest you post your 3dSeg/3dSkullStrip questions separately,
probably for Ziad to see. Note that posting an image showing what
you think is bad would be helpful.

- rick