Hi Michel,
> - apparently, the phase encode direction is considered as Y and deformation
> field and jacobian are calculated for this direction.
> Then, the defaults.realign.write.wrap value to indicate the
> direction of the deformation should be always set to [0 1 0]
> Is this correct ?
No, defaults.realign.write.wrap has no (well) relation to the direction
of the distortion. I am afraid Unwarp is a little poorly written in the
sense that the direction of distortions (phase-encode direction) is
hard-coded to be the y-direction (anterior-posterior).
The .wrap switch tells how to treat the case when (during realignment)
an object is moved outside of the image volume. When it is set to zero,
then in that direction the protruding part of the object is just
discarded. When it is set to one the object is wrapped back to the other
end of the FOV. This would intuitively seem really stupid, but it is
actually a good reflection of what happens in the phase-encode
direction.
> - How the static field values,if provided, are used during the estimation of partial
> derivatives of EPI deformation fields (spm_uw_estimate) ? It seems that there
> are only applied for the final calculation of images deformation
> (spm_get_image_def). is this right ?
It depends on what you mean by "final calculation". The partial
derivatives of the image w.r.t. to the DCT basis-functions are estimated
at the present sampling points, which will depend on the static field.
>
> -what does the comment "the possibility to include measured field-map
> temporally removed" in spm_realigned_ui exactly mean ?
Oops, I suspect it means that for a while (pre-release) it was
temporarily removed. I guess we should permanently remove that,
admittedly quite misleading, comment.
Good luck Jesper
|