Hi Ben,
Daniel Simmonds had a similar problem a while ago and used the robust
weighted least squares toolbox by Joern Diedrichsen. I also sometimes
use it and it seems like a very interesting alternative. The thread can
be found here
(www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0802&L=SPM&P=R6925).
Personally, I like to use more robust block designs in children, but
that of course depends on your questions.
Incidentally, does anyone know of an "accepted" approach to
disqualifying studies based on the amount of motion? As far as I could
gather, most people accept that motion exceeding one voxel size is very
difficult to correct, but there certainly is a detrimental effect of
less motion (1, 2 mm or so). And how do we judge rotations: is one
degree as bad as 1 mm ? Has anyone looked at that in a more systematic
fashion?
Best,
Marko
Ben Yerys schrieb:
> Hello--
>
> I have a pediatric dataset that is hampered by motion-related artifacts. I
> have tried using motion as a covariate, ArtRepair and FSL's melodic algorithim
> to remove artifacts, but have found inconsistent results (good for some
> participants but not others). I have also started to remove the individual
> scans that are artifact ridden and modeled them as a separate condition in my
> event-related design. This last solution appears to reduce the artifact
> activations more than my previous attempts. Despite the disadvantage of
> losing power in an event-related design, are there other reasons that speak
> against using this method for artifact correction?
>
> Thanks so much for your input,
> Ben
>
>
>
--
=====================================================================
Marko Wilke (Dr.med./M.D.)
[log in to unmask]
Universitäts-Kinderklinik University Children's Hospital
Abt. III (Neuropädiatrie) Dept. III (Pediatric neurology)
Hoppe-Seyler-Str. 1, D - 72076 Tübingen
Tel.: (+49) 07071 29-83416 Fax: (+49) 07071 29-5473
=====================================================================
|