Hi,
How long is it taking?
You might be able to make it run faster if you have an image
with a large FOV by cropping it to get rid of neck and any
unused space. However, in general it should take less than
an hour and if we had a way of running it faster, then we would
let everyone use that.
As for the cost function. You cannot use the warp field directly
with flirt but if you resample your image (same as the --iout
image generated by fnirt - not the warp/coeff files)
then you can feed in the reference and this resampled image
with no -init to the flirt call and it will calculate the cost function
between these two images with the same schedule.
All the best,
Mark
siamak wrote:
> Hi Mark,
>
> Thank you again for your very good comments. I am applying the fnirt
> command for the aim of nonrigid registration, I have the following problems:
>
> 1. This command is very time consuming, since I register only structural T1
> images, is there any config file that runs the command faster?
>
> 2. If I want to see the results of cost functions, I should use the same
> schedule file that you sent me or not.
>
> all the best
>
>
|