Hi -
Your seed is probably outside the brain. This sounds strange, but it
often happens due to subtle technical errors.
If you are running in simple mode, make sure you are specifying your
seeds in voxels (not mm) and that you have the mm option turned off.
If you are running in seedmask mode, make sure that your seed mask is
in diffusion space (i.e. same voxel sizes/dimensions as the diffusion
image)
Cheers
T
On 21 Nov 2006, at 11:58, Ed Roberts wrote:
> Hi, that's reassuring at least. I don't think it's running out of
> ram, I
> hope not, it's got 2GB to play with! I did change the threshold on BET
> because I was losing a significant amount of frontal lobe running at
> 0.5...
>
> I haven't currently got the structurals for running the registration
> program, but is it possible to test the data by running a single seed
> voxel on probtrack? I tried with a couple of voxel locations but it
> produces a file which was 4KB and blank in MRICRO, when I loaded it
> into
> FSLVIEW the program immediately closed - is this because there was
> nothing
> there and I should just keep looking?
>
> Thanks for the swift replies and patience.
>
> Ed
>
>
>> Hi - I think it has all worked OK -
>> 48 hours is a very long time for this data size though.
>>
>> The only things I can think of are that you either running out of
>> RAM,
>> or that your brain mask includes a lot of voxels that are not in the
>> brain.
>>
>> T
>>
>>
>>
>>
>> On 20 Nov 2006, at 17:45, Ed Roberts wrote:
>>
>>>> Hi there
>>>>
>>>>>
>>>>> I'm using 64 directions and 7 B0, my first question is; before
>>>>> converting
>>>>> 3D-4D to make the data file, should I have my B0 images all
>>>>> together at the
>>>>> start [no. 0-6], or should I average them and just use one.
>>>>>
>>>>
>>>> It's fine to have them all in, as long as the bvals and bvecs
>>>> match!
>>>>
>>>>> I've tried running Bedpost with all 71 images concatenated into
>>>>> the
>>>>> data
>>>>> file and it takes about 48 hours, although it's a 3T scan, it's
>>>>> running on a
>>>>> fast computer; this worries me.
>>>>
>>>> The Field strength will make no difference to the time taken. What
>>>> are the image dimensions?
>>>>
>>>>
>>>>>
>>>>> When Bedpost finished the log file says it has processed up to
>>>>> ".0059_finished" - Does that sound right?
>>>>
>>>> If there are 60 slices, yes.
>>>>>
>>>>> So finally, my idea is to run bedpost_datacheck, but I can't
>>>>> find a
>>>>> post
>>>>> which tells me how to do it.
>>>>>
>>>>
>>>> Bedpost_datacheck works on the input data not on the output data.
>>>>
>>>> you can call it by typing
>>>> bedpost_datacheck <inputdirectory>
>>>>
>>>>
>>>> T
>>>>
>>>
>>> Hi Tim,
>>>
>>> The dimensions of the images are 128 128 60, so i guess that solves
>>> the
>>> slices problem. The input data file is about 70mb in nii.gx format,
>>> does
>>> that explain why it takes so long?
>>>
>>> I ran the data check and I think it looks ok,
>>>
>>> $ bedpost_datacheck Tract
>>> Tract/data
>>> data_type INT16
>>> dim1 128
>>> dim2 128
>>> dim3 60
>>> dim4 71
>>> datatype 4
>>> pixdim1 1.8799999952
>>> pixdim2 1.8799999952
>>> pixdim3 2.5000000000
>>> pixdim4 1.0000000000
>>> cal_max 0.0000
>>> cal_min 0.0000
>>> file_type NIFTI-1+
>>>
>>> Tract/nodif
>>> data_type
>>> dim1 128
>>> dim2 128
>>> dim3 60
>>> dim4 1
>>> datatype 4
>>> pixdim1 1.8750000000
>>> pixdim2 1.8750000000
>>> pixdim3 2.5000000000
>>> pixdim4 0.0000000000
>>> cal_max 0.0000
>>> cal_min 0.0000
>>> glmax 2935
>>> glmin 0
>>> origin1 0
>>> origin2 0
>>> origin3 0
>>> file_type ANALYZE-7.5
>>>
>>> Tract/nodif_brain_mask
>>> data_type INT16
>>> dim1 128
>>> dim2 128
>>> dim3 60
>>> dim4 1
>>> datatype 4
>>> pixdim1 1.8750000000
>>> pixdim2 1.8750000000
>>> pixdim3 2.5000000000
>>> pixdim4 1.0000000000
>>> cal_max 1.0000
>>> cal_min 0.0000
>>> file_type NIFTI-1+
>>>
>>> num lines in Tract/bvals
>>> 1
>>> num words in Tract/bvals
>>> 71
>>> num lines in Tract/bvecs
>>> 3
>>> num words in Tract/bvecs
>>> 213
>>>
>>> The nodif is Analyze 7.5 instead of NIFTI - should this matter?
>>>
>>> Cheers,
>>>
>>> Ed
>>
|
