Dear Mark,
Thanks for this quick reply.
The only fsl command I could find was 'fugue', which runs after BBR.
It takes up around 98 to 100.5 % cpu time, it is using 38 Megabyte in
memory. I also attached a screenshot of top's output.
We acquired 59 slices for that subject on a 4 Tesla machine, but with no
unusually high resolution.
Maybe the problem could also be that as a T1 image I am using the SNR T1
image. Is the contrast-to-noise T1 image maybe much better to use?
Thank you,
Barbara
On 28.01.14 09:17, Mark Jenkinson wrote:
> Hi,
>
> This is far from normal, as usually epi_reg finishes in well under an hour. Do you have particularly high resolution images? The FAST segmentation of the structural image, that is needed by the BBR method used in epi_reg, can take a longer time to run if the image has very high resolution, although this long still sounds excessive unless the program is struggling to memory.
>
> If you run "top" from a terminal (or some form or activity monitor), do you see any processes that are taking up a lot of CPU time that are from FSL (e.g. fast, flirt, etc.)? And how much memory (RSIZE in top) are they using?
>
> All the best,
> Mark
>
>
> On 27 Jan 2014, at 16:06, Barbara Kreilkamp <[log in to unmask]> wrote:
>
>> Dear FSL-Users,
>>
>> As my epi_reg command has been running for over 16 hours, I would please like to ask:
>> Does anyone have experience with the approximate runtime of epi_reg when registering DTI to T1 data while also applying fieldmap correction as copied below?
>>
>> epi_reg --epi=epi_nodif_brain --t1=T1 --t1brain=T1_brain --out=nodif_brain_2_T1 --fmap=fieldmap --fmapmag=magnitude --fmapmagbrain=magnitude_brain --echospacing=0.000666667 --pedir=-y
>>
>> The command seems to take a lot of time especially after the BBR step (this happens on a Mac and a Linux machine).
>> Thanks a lot for your time and effort!
>>
>> Barbara
>>
>> University of Trento
>> Universität Osnabrück
|