[Camino-users] tractography on a normalized tensor data

[Philip A Cook]

dtlutgen uses a single orientation for the test function, meaning that it won't account for any variance in uncertainty caused by the placement of the diffusion gradients. So I think using the same scheme file will be no worse overall. You can run dtlutgen with different random seeds to get a different orientation for the test function. Doing this a few times and then averaging the results would smooth out the lookup table parameters, but might be overkill.

There's other issues with tracking in MNI space, particularly if there's a nonlinear transformation, in which case the reorientation of the tensors (and hence their relative orientation to the original gradients) would be variable anyway, also changes to the geometry of the tracts would likely influence results. Any changes in the anisotropy of the tensors due to interpolation will introduce differences in the estimated uncertainty.

I agree the best way would be to track in subject space using the Bayesian algorithm and transform the resulting lines.

On Aug 14, 2013, at 2:32 PM, Alexander, Daniel wrote:

> Hi,
> 
> Using the original schemefile is not perfect, but not unreasonable.  It is hard to think of a better way to do it if you really have to track in the MNI space.  The other option of course is to track in the native space and transform the streamlines into MNI space instead, which avoids some of these complications.
> 
> Danny
> 
> 
> On 9 Aug 2013, at 23:23, Binod Thapa Chhetry <bt2318 at columbia.edu> wrote:
> 
>> Hi all,
>> 
>> I want to run tractography with my tensors in a standard MNI space. I have normalized the fitted single tensor data(from dtifit) using DTI-TK. Now, I could do nifti2camino and bring my normalized tensor data(from DTI-TK) in nii to camino format. 
>> 
>> I was hoping if somebody could tell me what would be the best/appropriate input model for probabilistic tractography. Since, I have normalized tensors, I was thinking of PICO but then I only have schemefile for the tensor in native space, not in MNI. Can I use the same schemefile?Or, should I use a different model altogether, and how?
>> 
>> Hope this is not too confusing.
>> 
>> Looking forward for a reply.
>> Binod
>> 
>> 
>> _______________________________________________
>> Camino-users mailing list
>> Camino-users at www.nitrc.org
>> http://www.nitrc.org/mailman/listinfo/camino-users
> 
> 
> _______________________________________________
> Camino-users mailing list
> Camino-users at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/camino-users