Dear Daniel,
daniel weissman wrote:
> Dear SPMers,
>
> I was wondering whether anyone has any ideas about how much motion
> is too much motion. In a current study that we are conducting, we are
> collecting functional data with a voxel size of 3.75 X 3.75 X 5 mm. When
> realigning a subject's functional data from this study, is 2 mm in the x-,
> y-, or z- directions too much motion? What about 3 mm,or 10 mm? SPM
> produces a postscript file that indicates how much motion it has estimated
> during realignment, but would realignment work well if there was too much
> motion? Is there an amount of motion (e.g., 2 mm, 3 mm , 5mm) at which
> point realignment becomes less successful? Or, is it always equally
> successful?
>
I think you can say that realignment is always (almost) succesful in terms of
finding pretty correct estimates of the movement parameters.However, it is
clear that the realigned data are not identical to those you would have
collected had there been no motion (See Friston et al, MRM 96;35:346-355).
There are a number of tentative explanations for why we cannot recreate data
correctly including i) interpolation errors, ii) intra-scan movements and
iii) movement-by-suceptibility interactions. Of these error sources it is
mainly the latter which is expected to increase with increasing displacement
of the object. The magnitude of the errors due to this source will be highly
variable across the brain, and depend also on the strength of your magnet.
Even for very small movements these effects may be quite large in especially
the temporal lobes and the orbitofrontal cortex. It is therfore not easy to
give a general blanket endorsment of movement up to any one limit.
My, cautious, recommendation would be to include the realignment parameters
as confounds in the design matrix as soon as you have any appreciable
movement (>0.5mm). That way you will always err on the side of caution. Be
aware though that this may potentially decrease your senitivity to "true"
activations.
> Thanks in advance!
> :> Daniel Weissman
Good
luck Jesper
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|