The end effect of using motion estimates as covariates will depend on
the design matrix, specifically to extent to which such covariates
correlate with design regressors. We have a paper in press on this issue
(apologies for the self-promotion):
http://dx.doi.org/10.1002/hbm.20219
-Tom Johnstone
> Dear Katrin and Julia
>
> This procedure is used to account for residual movement artifacts. The
> procedure is described in a paper by Karl in 1996:
>
> @article{Friston96,
> Author = {Friston, K. J. and Williams, S. and Howard, R. and
> Franckowiak,R. S. J. and Turner, R.},
> Journal = {Magn Reson Med},
> Pages = {346-355},
> Title = {{Movement-Related effects in {f{MRI}} time series}},
> Volume = {35},
> Year = {1996}}
>
> My general impression is that the effects are always present.
>
>
> Hope this helps
> Torben
>
> Torben E. Lund
> Danish Research Centre for MR
> Copenhagen University Hospital
> Kettegaard Allé 30
> 2650 Hvidovre
> Denmark
> email: [log in to unmask]
> webpage: http://www.drcmr.dk
>
>
>
> > Dear SPM,
> >
> > once again a question concerning the use of the realignment
> parameters as
> > regressors in the design matrix:
> >
> > Some people first use these parameters to reduce differences between
> > certain volumes and the reference volume during preprocessing and
> then> include them again in the design matrix. But don´t you apply
> these> regressors on data that are changed according to these
> regressors in this
> > case?
> >
> > In other words: In the first step you estimate parameters that
> explain a
> > certain amount of variability in the data. Then you change the data
> > according to these parameters (reslicing the data). If you now
> include the
> > movement parameters in the design matrix they explain variance in
> data> that does not exist like that anymore.
> >
> > confused greetings,
> > Katrin and Julia
> >
> >
>
|