If you search the archives, you'll find that there isn't any
set-in-stone limit to the amount of acceptable motion. I've seen some
say that it shouldn't be larger than a voxel width/length, or some say
more than 1mm in any direction, etc.
I would suggest using Art Repair software by Paul Mazaika
(http://cibsr.stanford.edu/tools/ArtRepair/ArtRepair.htm). With this,
you can remove volumes from when the subject moved a lot. The
documentation here is quite good.
I believe Sue Whitfield-Gabrieli has some tools that can calculate
task-correlated motion.
You might as well run your analysis with motion parameters included. I
wouldn't trust the results you show here. That pattern around the
cerebellum looks quite suspect.
Also see these threads:
https://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind05&L=SPM&P=R387458&1=SPM&9=A&I=-3&J=on&d=No+Match%3BMatch%3BMatches&z=4
https://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind05&L=SPM&D=0&1=SPM&9=A&I=-3&J=on&d=No+Match%3BMatch%3BMatches&z=4&P=1921205
https://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind00&L=SPM&D=0&1=SPM&9=A&I=-3&J=on&d=No+Match%3BMatch%3BMatches&z=4&P=532003
Modestino, Edward J *HS wrote:
>
> Dear SPM community,
>
> According to the book “Functional Magnetic Resonance Imaging” (2004)
> by Huettel, Song and McCarthy, on page 267 “…head motion is often
> correlated with the experimental task, so including motion parameters
> in analysis models can also remove task-related activity. “ * How
> does one do this in SPM8? *
>
>
>
> In using realignment (motion correction), what is the acceptable range
> in mm’s? Is there a limit, say within 5 mm, which one should have
> within the data? If one is over that, does one not bother to use the
> data even if motion corrected?
>
>
>
> I analyzed the data using the moco files (motion corrected at
> scanner). The realignment (motion correction) in SPM 8 on this data
> divulged no more the 1.1 mm in any plane.
>
>
>
> I reran SPM8 using the regular EPI files (not moco). Here the range
> in realignment (motion correction) was between -2:+2 or 4 mm. After
> plotting this in MATLAB (via the rp_*.txt output), it does appear that
> the largest spikes are indeed task related.
>
>
>
> I have data that might show a motion related artifact in clusters at
> the stats level. Furthermore, this does appear to relate to the
> task. If I am motion correcting with the realignment, do I need to
> worry about this? Or should I assume the motion correction took care
> of this?
>
>
>
> The attached image shows the clusters at the stats level from data
> processed using the scanner motion corrected files (moco) and * then *
> running SPM8 realignment. Again, in the realignment process on this
> data divulged no more the 1.1 mm in any plane. * Does the attached
> image appear to show a motion related artifact in the sagittal view
> following the outline of the frontal, central, parietal, and occipital
> lobes and into the cerebellum? Or should I assume that this is actual
> activation as the realignment worked? Should I include motion
> parameters in the analysis models as this appears to be due to motion,
> or should I not bother with this as this is not due to motion artifact? *
>
>
>
> Thanks for your assistance,
>
> Ed Modestino
>
>
>
> Edward Justin Modestino, Ph.D.
>
> Postdoctoral Research Associate
>
> Ray Westphal Neuroimaging Laboratory
>
> Division of Perceptual Studies
>
> Department of Psychiatry and Neurobehavioral Sciences
>
> University of Virginia
>
> Email: [log in to unmask]
>
>
>
>
> ------------------------------------------------------------------------
>
|