(I apologize if anybody receives this twice; I had login trouble.)
All the diff_slice directories were created and the log files all indicate
that the slices were finished. When I originally cut and pasted the code
from the bedpost script, I got problems relating to bad inputs. So I
wrote my own script that doesn't use imglob in the avwmerge call, and this
produced a simple Segmentation Fault. So I ran avwmerge -z with only two
slices and this completed without error. Can I keep calling avwmerge -z
with a volume and a slice, one slice at a time, and thus gradually build
up the whole volume? I've previously been able to use avwmerge -t on
seven volumes without any problems. I have 1 GB of RAM.
Thank you for your reply,
Sam
> Dear Samuel
>
>> If I am not, please tell me where to go.
> ...
>
> We try to be a little bit more subtle than that .. :)
>
>
> It is poissible that what has happened here is that you don't have enough
> RAM to hold the full merged outputs. (Bedpost starts by processing every
> slice separately, then merges all the slices together).
>
> a couple of things to have a look for.
> 1) When it crashed, didi it leave behind all the processed slices in a
> directory called
> diff_slices
> in the bedpost directory
> if it did, you can just call avwmerge yourself, and test whether you go
> out of RAM (I don't know what your data look like, but you may well need
>> 200MB RAM to do this)
>
> However, it may have deleted this directory as part of the cleaning up
> process, depending on the crash behaviour.
>
>
> It is also worth looking at logs/log* in the bedpost directory
> These files will tell you whether each of the slices processed
> successfully.
>
> If they all did, then you know that you are dealing with a merging
problem
> at the end, and that this is likely to be because you are out of RAM.
>
>
> Note that you absolutely need 0.5GB RAM to run FDT, and prefarably you
> will have > 1GB
>
> Hope this is useful
>
> T
>
>
>
>
>>
>> I am using the "bedpost" program, part of FDT. The program ran for a
>> little more than a day, and according to the status messages it went
>> through all slices without trouble. Then, in the postprocessing phase,
>> it
>> produced a series of error messages and finished without constructing
>> all
>> the desired outputs. Since then I have run bedpost again, using the
>> same
>> directory as before, and each time the program ends with the same
>> results. These subsequent runs take only a few minutes, probably
>> because
>> it is using the processed slices from the original run.
>>
>> I am using FSL on Cygwin. The original error message I received was
>> immediately preceded by a notification from Windows that I was low on
>> virtual memory and that the amount of virtual memory would be increased.
>> This has not happened again since then.
>>
>> It strikes me as odd that some fatal error would happen in the
>> postprocessing phase, since, according to my reading of the bedpost
>> script, the only things happening there are some calls to avwmerge and
>> avwmaths.
>>
>> I noticed when I was reading through the help pages before posting to
>> this
>> email list that FSL provides a function (avwmerge) to build 4D image
>> files. My original input included a 4D ANALYZE file created from
>> another
>> MRI data program. I constructed it with the slices from the different
>> volumes (this being dtMRI, there is one volume for each angle)
>> interleaved. I'm wondering if FDT prefers non-interleaved data, so I
>> created a new file from the volumes with avwmerge and placed it in a new
>> directory and am currently running bedpost all over again on this new
>> directory.
>>
>> I checked my inputs beforehand with fslview and bedpost_datacheck, and
>> did
>> not find any problems.
>>
>> Attached with this email is a text file containing the error messages,
>> as
>> well as the output from fslerrorreport. I will appreciate any reply, as
>> my graduation is contingent upon this project.
>>
>> Thank you for your time,
>> Samuel M. Johnston
>> UC Berkeley
>>
>
> --
> -------------------------------------------------------------------------
------
> Tim Behrens
> Centre for Functional MRI of the Brain
> The John Radcliffe Hospital
> Headley Way Oxford OX3 9DU
> Oxford University
> Work 01865 222782
> Mobile 07980 884537
> -------------------------------------------------------------------------
------
>
|