Hi Everyone,
I have some philosophical questions about motion for whomever would be up to
providing input...
Some threads in the past have suggested that for brief and sudden head motion of
subjects, removing the volume and replacing it with an average of the volumes on
either side would be helpful. My questions are:
1. How many (or what percentage) of the volumes can one remove without
autocorrelation becoming a serious problem.
2. I've noticed that several individual brief head motions sometimes does more to
make an image look messy than total movement across the scan. I would like to be
consistent across my subjects. Does anyone have a rule of thumb about which
volumes they remove (i.e. relative movements greater than a certain size) that is a
good compromise between keeping data and excluding brain analyses that are
obviously full of motion artifact.
3. If the head motion is correlated with the task (as determined by running a
correlation between the relative motion in the mc file and the paradigm), is there a
threshold of motion below which that is not an issue and data can still be included,
or should subjects for whom p<.05 for such a correlation always be excluded? For
instance, if in-plane resolution is 2mm, relative motion is .04 and absolute motion is
.15, it would seem as though correlation with paradigm may not even make a
difference. Is that the case?
Any guidance people could provide regarding any of these lengthy essay questions
(either through personal experience or existing literature) would be greatly
appreciated!
Thanks,
Meredith
|