Hi,
> 1.what is the actual difference between using the following commands:
> -u unwarp
> -w warp
> --unwarpdir
> It seems defining the direction of --unwarpdir=y and --unwarpdir=y-
> would be
> the equivalent of -u and -w. Is this a correct assumption?
No.
The unwarp option tries to undo the distortion in the image.
The warp option takes an initially undistorted image and create
distortion.
The unwarpdir is the direction that the unwarping will be done in.
The reason that unwarping in the opposite direction as it should be done
is not the same as warping is a subtle one and related to the fact
that the
distortion is represented by the fieldmap in an undistorted space. This
means that it matters whether you start with a distorted or undistorted
image for looking up what shift to apply. Don't worry if you don't
quite
follow this, as the main thing is that to undistort an image you want
the
unwarp option.
> 2. Is there an advantage of using the command --phaseconj?
No. In fact in my tests I have found this to perform worse.
> 3. Is it necessary to warp the phasemap to match the distortion of
> the EPI
> images before using it to unwarp/warp the EPI images?
The EPI and fieldmap need to be registered, so we use the warp
option, applied to the fieldmap magnitude image, to achieve this.
However, all of this is automatically taken care of if you do your
unwarping via the pre-stats part of the Feat GUI. I strongly recommend
that you use this as it applies several useful steps. You can setup
the GUI to only perform pre-stats (where you turn everything off
except the unwarping) and so there is no extra overhead either.
> 4. Is there any published information pertaining to the HBM 2001,
> Jenkinson,
> which was referenced in an earlier posting?
We do not have a specific paper relating to this implementation as
it is based on existing methods - see Cusack et al, Neuroimage,
18(1):127-42, 2003, for details, especially with respect to the
registration.
All the best,
Mark
|