I am scanning patients with movement disorders and for some of them I’d like to combine (2 or 3) structural T1 images to reduce motion-related artifacts. To my surprise, google could not find me a quick answer on how to do that in the most optimal way.
Can I simply co-register and average them? Or is there a better way/tool to do this?
indeed you can coregister with rigid body registration and average. however, depending on the duration of time between the images this may not be ideal. in freesurfer, all input images are registered to create an average image to improve SNR and they also have a longitudinal processing stream that can take into account temporal differences in structure.
if you have significant motion in some images, but not others, depending on the amount of motion, it may be better to toss out the images rather than averaging.
Thanks so much!
I also realize now that fmriprep allows multiple T1w’s. I’ll do some test runs with and without the worst images to see what works best.