Hi -
Matthew Brett wrote:
> I did a very tiny study of including movement parameters up to the 24
> regressor spin-history model (which will include the effects modeled
> by the difference of the parameters) and found, like Tom Johnstone,
> that only the movement parameters themselves seemed to be robustly
> useful:
So the most important choice is between
a. including the 6 motion parameters for each scan as regressors in the
design matrix; or
b. computing the displacement for every voxel co-ordinate (by
transforming voxel (i,j,k) to (x,y,z) and then compute the L2-norm
between (i,j,k) and (x,y,z) and use that to correct for artifacts.
Have these two ever been compared?
It is also possible to apply PCA to the motion parameters (to have fewer
regressors) - in order not lose unnecessarily much power, see
http://www-bmu.psychiatry.cam.ac.uk/sitewide/publications/journal/bullmore99met.pdf.
The PCA method has been applied and yielded nice results, not sure if
this was ever published.
Best wishes,
Alle Meije Wink
