Ok, so this way the NxN self-connectivity tracts of a mesh will be
separated into
#N x (1xN) tractograms (seeding from each vertex to whole mesh as a target)?
Unfortunately if I want to feed a thresholded matrix1_mask into this
pipeline in order to get the tracts-of-interest, I will also need to
separate tractograms based on every seed vertex AND target vertex,
i.e. #(NxN) x (1x1). Then I can read matrix1_mask, find the
corresponding tractogram in #(NxN) and add up all.
(Probably I haven't managed to put my problem into correct words,
literally speaking I only want to "feed" a matrix1_mask and associated
cortical surface to probtrackx2, and get the tractogram which
"creates" this connectivity-of-interest.)
Thanks for your time,
Sourena
On 6/15/12, Saad Jbabdi <[log in to unmask]> wrote:
> Ok, then what you are after is to get "separate" tractography results for
> each of your seed points of interest in one probtrackx2 call.
> This is now possible using the --omatrix2 option. If you set the --target2
> entry to a brain mask in seed space, then the matrix2 result will contain
> all your tractography result separate for each seed. The rows of the matrix
> will correspond to the spatial histogram of tracts for each seed.
> Then if you want to add up tracts from different seeds, you just sum up
> rows.
>
>
> Saad
>
> On 14 Jun 2012, at 16:20, Sourena Soheili wrote:
>
>> Hi
>> From --help command I understand that in --network mode any path which
>> touches at least one of the targets will survive. Now suppose we have
>> a segmented matrix1 which only encodes bilateral portions of the
>> corticospinal tract for hand movement, meaning that it only includes:
>> tract1: brainstem vertices to right hand motor cortex vertices
>> tract2: brainstem vertices to left hand motor cortex vertices
>> If I run in --network mode, won't I get some extra tracts connecting
>> right and left motor cortices?
>> (Please consider that the real scenario can be much more complicated,
>> and there could be no several "discrete" mesh patches to seed to/from,
>> rather, a single confluent mixure of thresholded matrix1 entries.)
>> What can be the best approach here?
>>
>> Cheers,
>> Sourena
>>
>> On 6/14/12, Saad Jbabdi <[log in to unmask]> wrote:
>>> Hi - you can do this by running probtrackx2 with the --network option
>>> on,
>>> and regions A and B as two separate seed masks/surfaces. These should be
>>> entered as a text file that contains paths to the relevant files. The
>>> result
>>> will be the connection maps from A to B and from B to A added together.
>>>
>>> Cheers
>>> Saad
>>>
>>> On 14 Jun 2012, at 13:04, Sourena Soheili wrote:
>>>
>>>> Dear Saad
>>>> Suppose we have a disease condition which disrupts connectivity of
>>>> cortical region A to B. We run probtrackx with Matrix1 to record
>>>> repertoire of cortico-cortical connectivity information in patients
>>>> group as well as healthy controls, and run t-statistics. So the
>>>> omatrix1 is converted to a statistical significance map with the same
>>>> dimensionality.
>>>> This map only encodes cortical regions A and B (as thresholded
>>>> significant mesh vertices), not the underlying tracts which are
>>>> influenced by the disease condition.
>>>> So I am seeking a solution to visualize the tracts which show a
>>>> disease-related impact.
>>>>
>>>> I wonder what is the best approach to span over all streamlines like
>>>> this:
>>>>
>>>> For each #voxel in Significant_Matrix_1; Do
>>>> Let foo = seed voxel associated to #voxel
>>>> Let bar = target voxel associated to #voxel
>>>> Run probtrackx with options seed=foo target=bar
>>>> Done
>>>> Sum up all tractograms associated with NxN vertex connections
>>>>
>>>>
>>>> On 6/14/12, Saad Jbabdi <[log in to unmask]> wrote:
>>>>> Not sure I follow you. I guess you want to track from specific
>>>>> vertices
>>>>> (depending on the results of some analysis).
>>>>> You can do that by setting the scalar values of your seed surface to
>>>>> 1's
>>>>> and
>>>>> 0's.
>>>>>
>>>>> Saad
>>>>>
>>>>> On 13 Jun 2012, at 22:53, Sourena Soheili wrote:
>>>>>
>>>>>> Hi Saad
>>>>>> Consider some manipulation is performed on raw matrix1 (e.g.
>>>>>> groupwise
>>>>>> comparisons and zeroing of the non-significant connections, or a
>>>>>> simple segmentation). Is there a method to feed the manipulated
>>>>>> matrix1 into the probtrackx pipeline and visualize the tractogram
>>>>>> which renders only the significant connections? (The only option
>>>>>> which
>>>>>> comes to my mind is vertexwise reading of connections-of-interest
>>>>>> from
>>>>>> manipulated matrix1 and then performing separate vertex-to-vertex
>>>>>> tractographies and lastly summing up all the results.)
>>>>>>
>>>>>> Is there a better option available?
>>>>>>
>>>>>> Cheers,
>>>>>> Sourena
>>>>>>
>>>>>> On 6/12/12, Saad Jbabdi <[log in to unmask]> wrote:
>>>>>>> Yes.
>>>>>>>
>>>>>>>
>>>>>>> On 12 Jun 2012, at 14:19, Sourena Soheili wrote:
>>>>>>>
>>>>>>>> Hi Saad
>>>>>>>> Is the NxN matrix arranged based on vertex numbers (as extracted
>>>>>>>> from
>>>>>>>> the source mesh topology file) in every dimension?
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> Sourena
>>>>>>>>
>>>>>>>>
>>>>>>>> On 6/12/12, Saad Jbabdi <[log in to unmask]> wrote:
>>>>>>>>> Hi
>>>>>>>>> If the seed mask is a surface omatrix1 will encode
>>>>>>>>> vertex-to-vertex
>>>>>>>>> connectivity of the seed mask. Dimensions are NxN where N is the
>>>>>>>>> number
>>>>>>>>> of
>>>>>>>>> vertices.
>>>>>>>>> Detection of surface crossing is done using triangles, then the
>>>>>>>>> nearest
>>>>>>>>> vertex within the crossed triangle is considered for counting.
>>>>>>>>>
>>>>>>>>> Cheers
>>>>>>>>> Saad
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On 11 Jun 2012, at 20:21, Sourena Soheili wrote:
>>>>>>>>>
>>>>>>>>>> Hi,
>>>>>>>>>> I have a few questions about probtrackx2 and --omatrix1 option,
>>>>>>>>>> If a triangular freesurfer surface is used as a seed mask, does
>>>>>>>>>> the
>>>>>>>>>> omatrix1 encode "triangle" to "triangle" surface connectivity
>>>>>>>>>> information?
>>>>>>>>>> (I've seen in a previous post that it is not a vertex-wise
>>>>>>>>>> matrix)
>>>>>>>>>>
>>>>>>>>>> What will be the dimensionality of the omatrix1 nifti file?
>>>>>>>>>>
>>>>>>>>>> Is it correct that the seed/target voxels in each nifti dimension
>>>>>>>>>> are
>>>>>>>>>> arranged and ordered by their #face or #vertex in the input mesh?
>>>>>>>>>> Thanks in advance,
>>>>>>>>>> Sourena Soheili
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> --
>>>>>>>>> Saad Jbabdi
>>>>>>>>> University of Oxford, FMRIB Centre
>>>>>>>>>
>>>>>>>>> JR Hospital, Headington, OX3 9DU, UK
>>>>>>>>> (+44)1865-222466 (fax 717)
>>>>>>>>> www.fmrib.ox.ac.uk/~saad
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>> Saad Jbabdi
>>>>>>> University of Oxford, FMRIB Centre
>>>>>>>
>>>>>>> JR Hospital, Headington, OX3 9DU, UK
>>>>>>> (+44)1865-222466 (fax 717)
>>>>>>> www.fmrib.ox.ac.uk/~saad
>>>>>>>
>>>>>>
>>>>>
>>>>> --
>>>>> Saad Jbabdi
>>>>> University of Oxford, FMRIB Centre
>>>>>
>>>>> JR Hospital, Headington, OX3 9DU, UK
>>>>> (+44)1865-222466 (fax 717)
>>>>> www.fmrib.ox.ac.uk/~saad
>>>>>
>>>>
>>>
>>> --
>>> Saad Jbabdi
>>> University of Oxford, FMRIB Centre
>>>
>>> JR Hospital, Headington, OX3 9DU, UK
>>> (+44)1865-222466 (fax 717)
>>> www.fmrib.ox.ac.uk/~saad
>>>
>>
>
> --
> Saad Jbabdi
> University of Oxford, FMRIB Centre
>
> JR Hospital, Headington, OX3 9DU, UK
> (+44)1865-222466 (fax 717)
> www.fmrib.ox.ac.uk/~saad
>
|