afni.nimh.nih.gov

Why don't you read the -help output for 3dvolreg carefully and figure out what you need? Or try different options and see if the results are better?

In the case where the images are acquired in the sagittal plane, and the subject's dominant head movement is a nodding motion, then a number of people have reported significant improvement by using 2dImReg followed by 3dvolreg.

bob cox