[Camino-users] answers to recent questions about tractography

[Peter Neher]

Hi Philip, thanks for the explanations!

On 07/12/2012 04:39 PM, Philip A Cook wrote:
> Hi all,
>
> Batching up these answers
>
> On Jul 11, 2012, at 3:14 PM, Curci Marina Carmela wrote:
>
>> I have a question: if I run the track command and then procstreamlines using as input for track the pds file, I have noticed that the number of streamlines generated with track is different from the number of streamlines processed by procstreamlines. Could you explain to me the reason, please??
> You should get one streamline per principal direction in a voxel, one streamline will be initiated along each of the principal directions. The total number of streamlines will therefore usually be greater than the number of seed points, unless you only have one PD per voxel.
>
>
> On Jul 12, 2012, at 3:18 AM, Peter Neher wrote:
>
>> So do you proceed along the streamline with a fixed step size, or do you somehow calculate the point where the streamline is leaving/entering a voxel explicitely in order to find the point where to apply the new direction?
> FACT calculates the point of entry into the next voxel. In a future release I will add an option for fixed step size with nearest-neighbor interpolation.
>
>
> On Jul 12, 2012, at 3:23 AM, Peter Neher wrote:
>
>> And here comes the next question. Is it normal for the bayesian and wild bootstrap tracking algorithms to yield fiber files for a whole brain of about 15Gb? Not easy to handle :)
>> I guess this result contains for example for 1000 bootstrap iterations, 1000 fibers per seedpoint. Shouldn't these 100 fibers somehow be combined to one final resulting fiber?
>> And another question. Is there some way to define a minimum fiber length for the resulting fibers?
>
> The fibers from each voxel are not combined into a single output fiber, hence the large files. You can combine them into images in various ways.
>
> You can throw out short tracts with procstreamlines -mintractlength X, where X is in mm.
>
>
>
> On Jul 11, 2012, at 9:10 AM, Peter Neher wrote:
>
>> Hi everybody,
>>
>> I was just using camino FACT tracking on a simple phantom dataset using the following command:
>>
>> track -inputmodel dt -seedfile "mask.nii" -anisthresh 0.1 -ipthresh 0.7 -inputfile dti.Bdouble > factDtiTracts.Bfloat
>>
>> I also tried "-curvethresh 45" but the result still lokks like no angular threshold is applied. Even if I tried -ipthresh 1.0 which should result in no allowed curvature at all, the result does not really change. A screenshot of the 1.0 - result is attached.
>
> The picture shows the pdview RGB / PD map of the synthetic data in question.
>
> I think there are two issues here. The curvature is evaluated periodically, and the period is set by the slice dimension of the data, in this case 7mm. The idea is to avoid making the curvature threshold dependent on step size / use of interpolation. This means tracts might continue a bit further before terminating in anisotropic data such as this (2x2x7mm). The other reason that you see sharply curved paths is that when you seed tracts at every voxel, you include the inflection point of the curves. When tractography is seeded, fibers are tracked independently in two directions, forwards (along the principal direction) and backwards (in the opposite direction), and the results are then joined. Curvature is evaluated independently for each segment.
>
> I tried tracking only at the vertical ends of the crossing (the top and bottom green voxels), and I got the expected results, so I think the curvature threshold is working normally. I've added a note to the track man page to clarify this.
>   

-- 
Dipl.-Inform. Peter Neher
German Cancer Research Centre
(DKFZ, Deutsches Krebsforschungszentrum in der Helmholtz-Gemeinschaft, Stiftung des öffentlichen Rechts)
Division of Medical and Biological Informatics
Im Neuenheimer Feld 280, D-69120 Heidelberg

Phone: 49-(0)6221-42-3552, Fax: 49-(0)6221-42-2345
E-Mail: p.neher at dkfz-heidelberg.de, Web: www.dkfz.de

The information contained in this message may be confidential and legally protected under applicable law. The message is intended solely for the addressee(s). If you are not the intended recipient, you are hereby notified that any use, forwarding, dissemination, or reproduction of this message is strictly prohibited and may be unlawful. If you are not the intended recipient, please contact the sender by return e-mail and destroy all copies of the original message.