Hi Erik,
It is possible to make a specific schedule file to do just this.
I have several candidate schedule files in $FSLDIR/etc/flirtsch
but admittedly am not giving tutorials in how to write schedule
files. There are ones that are close in this directory, but none
that do precisely what you want. However, you are welcome
to modify one and see if you can get it to do what you want.
It is mainly the paramsubset that you need to change.
It is arranged as a 12xN matrix where N is the number of
parameters you want to modify (3 for you) and the order
of the parameters is: Rx Ry Rz Tx Ty Tz Sx Sy Sz Kx Ky Kz
where R=rotation, T=translation, S=scaling, K=skew
If you take say sch2D_3dof and put the ones in the parameters
that you want to modify, then you'll be a large part of the way
there.
As for timing though, I'm afraid the number of parameters does not
have a huge impact on the timing. A lot of time is spent doing
searches and perturbations which you will still have. You could
turn them off to really increase speed, but then you risk a lack of
robustness. Given that it seems to take about 4 minutes per volume,
that seems quite reasonable to me. I'm afraid it is a question of
speed versus quality/robustness. I tend to go for the latter.
Hope this helps.
All the best,
Mark
On Wednesday, May 7, 2003, at 02:24 pm, Erik-Jan Vlieger wrote:
> Hi,
>
> I have been using flirt in a 6DOF-2D mode to perform eddy current
> corrections,
> and this works very well. However, the expected distortions can be
> modeled
> using less degrees of freedom:
> - a translation along the phase encoding direction
> - a shear parallel to the phase encoding direction
> - a scaling in the phase encoding direction
>
> Is it possible to instruct flirt to search only along these variables?
> Could
> one use the schedule files for that? I assume it would save time
> (currently,
> on a 1.6 Ghz pentium, the correction of one volume of 16 128x128
> slices,
> with 4 acquisitions of 6 directions of diffusioin weighting, takes 1.5
> hour...)
>
> Thanks a lot,
>
> Erik-Jan Vlieger
> --
> E.J.P. Vlieger
> Academic Medical Center
> Dept of Radiology, G1 - 209
> Meibergdreef 9
> 1105 AZ Amsterdam
> The Netherlands
> Tel: +31 20 5664975 / 5669111 pager 63539
> Fax: +31 20 5669119
> Email: [log in to unmask]
> Home: [log in to unmask]
|