In addition to John's feedback I would alsomsuggest looking at the data as a movie with something like fslview. This helps see when motion is "bad." What I meN by bad is that sometimes motion causes big MR artifacts where full slices will be black or at least dark. In my opinion this helps decide whether motion is too big.
Jason
Dear Michael and others,
It seems, that there is no simple “rule-of-thumb” to deal with this problem.
For block designs, inclusion of motion covariates has a deleterious impact on GLM sensitivity when activity of interest is correlated with motion, e.g., in tasks required manual action, speech-production or sudden sound resentation (Bullmore et al., 1999; Johnstone et al., 2006). Analysis using movement corrections and regressors should increase the significance and extent of activations detected before correcting for movement only if they are orthogonal to movement effects. If any component of activation or task-related changes are correlated with an estimated movement effect, any manipulation to minimize motion artifacts might lead to loss of sensitivity and removal of this component (Bullmore et al., 1999; Freire & Mangin, 2001; Friston et al., 1996; Hajnal et al., 1994).
Some recent papers dealing with this problem:
Churchill et al. - 2012 - Optimizing preprocessing and analysis pipelines for single-subject fMRI. I. Standard temporal motion and physiological noise correction methods.
Hutton et al. - 2013 - Phase informed model for motion and susceptibility
Power et al. - 2014 - Methods to detect, characterize, and remove motion artifact in resting state fMRI
As far as I understand, this is still unresolved issue and the answer is “It depends…”
Best,
Ella
On Fri, Mar 28, 2014 at 12:01 PM, Mouthon Michaël <[log in to unmask]> wrote:
Dear John, thank you for your last post.
But I have a question regarding including motion parameter of realignment as covariate in the GLM. The paper of Johnstone et al 2006 (http://www.ncbi.nlm.nih.gov/pubmed/16456818) shows that the inclusion of motion parameter as covariate had a deleterious impact on the GLM sensitivity for a fMRI block design and increase of sensitivity for an event-related design.
Do you think that it's a bad idea to include these motion parameters as covariate in any case for a block design ? Or it could help in the case of a population with a lot of movement like older people or patients even if it a block design?
Thank you in advance
Michaël
--Ella GabitovPh.D. studentThe E.J. Safra Brain Research Center for Learning DisabilitiesThe Laboratory of Human Brain and LearningUniversity of HaifaIsraelPhone: +972-52-6722700E-mail: [log in to unmask]