Hi,
PACE data per se (i.e. the non-moco EPIs) are corrected prospectively in that the positioning of the current volume is corrected for the movement detected between the previous and 'second but previous' volume recorded.
Thus, there is still movement in the data and it does not prevent you from applying unwarp. Also, I don't think you would want to supply the estimated PACE parameters to unwarp. However, these are stored in a temporary textfile and you could access them.
Cheers-
Andreas
________________________________________
Von: FSL - FMRIB's Software Library [[log in to unmask]] im Auftrag von Jesper Andersson [[log in to unmask]]
Gesendet: Freitag, 25. September 2009 17:19
An: [log in to unmask]
Betreff: Re: [FSL] Recommendations for Jesper's unwarping with PACE
Hi Joe,
> If I recall correctly, for dealing with head motion, you recommend
> that people either use your realign & unwarp tool or add estimated
> motion parameters into the statistical model as covariates, but not
> both. Is that correct?
>
> Do you have any recommendation if the data were acquired with
> prospective motion correction (PACE on Siemen's machines)? This
> came up the other day when I learned that one of my colleagues was
> acquiring her EPI data using PACE, then using your realign and
> unwarp code, and then including the residual motion estimates in the
> model. Seemed like overkill but I thought it worth getting a more
> informed opinion.
In general Unwarp (an SPM thingie that I worked on prior to seeing the
light) will remove a subset of the variance that gets removed when
including the movement parameters in the model. Its purpose is to try
an get rid of the "unwanted" residual movement correlated variance
while leaving any true activations (even if they are correlated with
the movement). Hence, if one includes the movement parameters in the
model there is no point in running Unwarp.
I am also not sure how well it would work together with PACE (if one
wanted to run Unwarp and _not_ include the movement parameters).
Unwarp uses the estimated movement parameters and try to work out how
much the magnetic field would change as a consequence of those
movements. If PACE has already detected/corrected for the movements
there will be little/nothing left for Unwarp to work with. Ideally
Unwarp should then have access to the movement parameters detected by
PACE, and these should be added to those estimated by Unwarp. There is
no mechanism for doing that though.
Good Luck Jesper
|