There is one other possibility I just spotted: Each sample streamline
runs in two opposite directions from a seed voxel. If one direction
leads to one target, and the other to another target, then you will
have a seed voxel connecting to two targets (despite your termination
mask), producing eventually a sum >5000 if this happens often...
Cheers,
Saad.
On 6 Apr 2009, at 15:35, Catherine Traynor wrote:
> I tried that and there was no overlap between the two regions.
>
> Is there any other reason why I might get a sum of >5000?
>
> Thanks, Catherine
>
>
> On 6/4/09 15:16, "Saad Jbabdi" <[log in to unmask]> wrote:
>
>> That's a possibility that you could check by transforming your masks
>> into diffusion space (using flirt -applyxfm -init <struct2diff> -
>> interp nearestneighbour)
>> If your target masks then overlap in diffusion space, it is possible
>> that a voxel from your seed mask connects to both, which would give
>> you the values >5000.
>>
>> Cheers,
>> Saad.
>>
>>
>> On 6 Apr 2009, at 15:04, Catherine Traynor wrote:
>>
>>> Hi Saad
>>>
>>> Having included a termination mask as well as the classification
>>> masks, if I
>>> sum the results from probtrack (the seeds_to* images) I get a value
>>> of >5000
>>> in some voxels, despite having set the number of samples to 5000.
>>>
>>> I was wondering if this should be happening, and if not, then why?
>>> (could my
>>> classification target masks be overlapping in diffusion space???)
>>>
>>> Any suggestions?
>>>
>>> Thanks again, Catherine
>>>
>>>
>>>
>>> On 6/4/09 14:40, "Saad Jbabdi" <[log in to unmask]> wrote:
>>>
>>>> Hi Catherine,
>>>> Probtrackx uses probabilistic interpolation (as described in
>>>> Behrens
>>>> et al, MRM 2003).
>>>> It basically takes a sample orientation from an adjacent voxel,
>>>> being
>>>> more likely to pick this sample from a voxel that is nearby.
>>>>
>>>> Cheers,
>>>> Saad.
>>>>
>>>>
>>>> On 6 Apr 2009, at 12:55, Catherine Traynor wrote:
>>>>
>>>>> Hi again Saad
>>>>>
>>>>> If I input my seed and target masks in structural space, what type
>>>>> of
>>>>> interpolation is used when these are transformed into diffusion
>>>>> space?
>>>>>
>>>>> Thanks, Catherine
>>>>>
>>>>>
>>>>> On 25/2/09 12:49, "Saad Jbabdi" <[log in to unmask]> wrote:
>>>>>
>>>>>> yes that's right.
>>>>>>
>>>>>> Saad.
>>>>>>
>>>>>>
>>>>>> On 25 Feb 2009, at 10:09, Catherine Traynor wrote:
>>>>>>
>>>>>>> Hi Saad
>>>>>>>
>>>>>>> So I should use the termination mask in combination with my
>>>>>>> classification
>>>>>>> targets?
>>>>>>>
>>>>>>> Thanks, Catherine
>>>>>>>
>>>>>>> On 25/2/09 09:42, "Saad Jbabdi" <[log in to unmask]> wrote:
>>>>>>>
>>>>>>>> Hi Catherine,
>>>>>>>>
>>>>>>>> You can use the --stop option (or termination masks in the
>>>>>>>> GUI). If
>>>>>>>> you merge both targets into a single mask (e.g. with fslmaths)
>>>>>>>> that
>>>>>>>> you use as a termination mask, then sample tracts that get to
>>>>>>>> any
>>>>>>>> of
>>>>>>>> these targets will terminate there.
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>> Saad.
>>>>>>>>
>>>>>>>>
>>>>>>>> On 24 Feb 2009, at 12:26, Catherine Traynor wrote:
>>>>>>>>
>>>>>>>>> I am carrying out the connectivity-based seed classification
>>>>>>>>> using
>>>>>>>>> two
>>>>>>>>> adjacent cortical target masks. I am concerned about paths
>>>>>>>>> that
>>>>>>>>> appear to
>>>>>>>>> travel through one of the target regions and on to the other.
>>>>>>>>> The
>>>>>>>>> seed voxel
>>>>>>>>> is then classified as connecting to both target regions for a
>>>>>>>>> single
>>>>>>>>> path.
>>>>>>>>>
>>>>>>>>> What is the best way to deal with this situation? Is there any
>>>>>>>>> way
>>>>>>>>> to
>>>>>>>>> terminate the path after it encounters the first target
>>>>>>>>> region,
>>>>>>>>> without
>>>>>>>>> having to run probtrack separately for each region?
>>>>>>>>>
>>>>>>>>> Many thanks
>>>>>>>>>
>>>>>>>>> Catherine Traynor
>>>>>>>>>
>>>>>>>>
>>>>>>>> Saad Jbabdi
>>>>>>>> Oxford University FMRIB Centre
>>>>>>>>
>>>>>>>> JR Hospital, Headington, OX3 9DU, UK
>>>>>>>> +44 (0) 1865 222545 (fax 717)
>>>>>>>> www.fmrib.ox.ac.uk/~saad
>>>>>>>
>>>>>>
>>>>>> Saad Jbabdi
>>>>>> Oxford University FMRIB Centre
>>>>>>
>>>>>> JR Hospital, Headington, OX3 9DU, UK
>>>>>> +44 (0) 1865 222545 (fax 717)
>>>>>> www.fmrib.ox.ac.uk/~saad
>>>>>
>>>>
>>>> Saad Jbabdi
>>>> Oxford University FMRIB Centre
>>>>
>>>> JR Hospital, Headington, OX3 9DU, UK
>>>> +44 (0) 1865 222545 (fax 717)
>>>> www.fmrib.ox.ac.uk/~saad
>>>
>>
>> Saad Jbabdi
>> Oxford University FMRIB Centre
>>
>> JR Hospital, Headington, OX3 9DU, UK
>> +44 (0) 1865 222545 (fax 717)
>> www.fmrib.ox.ac.uk/~saad
>
Saad Jbabdi
Oxford University FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
+44 (0) 1865 222545 (fax 717)
www.fmrib.ox.ac.uk/~saad
|