#  UKF tractography vs MRtrix CSD in my data 

**URL:** <https://discourse.slicer.org/t/ukf-tractography-vs-mrtrix-csd-in-my-data/33165>\
**Category:** SlicerDMRI\
**Created:** [December 1, 2023, 4:50pm UTC](https://discourse.slicer.org/t/ukf-tractography-vs-mrtrix-csd-in-my-data/33165 "2023-12-01T16:50:54Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![gaunny](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/gaunny/32/67838_2.png) [@gaunny](https://discourse.slicer.org/u/gaunny)\
**Post date:** [December 1, 2023, 4:50pm UTC](https://discourse.slicer.org/t/ukf-tractography-vs-mrtrix-csd-in-my-data/33165/1 "2023-12-01T16:50:54Z")

</div>

Hi!  
I meet problem when I use UKF tractography in my data. I used the default parameters in Slicer 3D, but the traced out fibres are very discontinuous and even in the wrong direction. For example, the corticospinal tract, pictured below, is broken.

 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/6/e/6ee01aa0e8d4d6debd30bbd96f9d2934ea2dd482.jpeg)  
I checked my data, it has 3 b-values, 0,1000 and 2000, and each b-value contains 9, 50, 50 directions. A slice of b0 is shown below:  
 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/b/a/ba969dd5d4cb4bc11005a5a71490606407071eaa.jpeg)  
I checked different slices in the same direction, and images in different directions of the same slice, and all were fine.  
When I use CSD in MRtrix, it works well, like this:  
 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/a/b/abdc6cfd7fa7b25b8f161f9a396ce85a1e0fd9ff.jpeg)

So I’m wondering why the UKF method can’t achieve proper tracking on this data?  
Looking forward to your reply.  
Best regards.

---

<div class="post-metadata">

**Author:** ![pieper](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/pieper/32/8_2.png) [@pieper](https://discourse.slicer.org/u/pieper)\
**Post date:** [December 1, 2023, 6:40pm UTC](https://discourse.slicer.org/t/ukf-tractography-vs-mrtrix-csd-in-my-data/33165/2 "2023-12-01T18:40:50Z")

</div>

Probably something in the gradient directions or measurement frame is incorrect.

---

<div class="post-metadata">

**Author:** ![gaunny](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/gaunny/32/67838_2.png) [@gaunny](https://discourse.slicer.org/u/gaunny)\
**Post date:** [December 5, 2023, 12:55am UTC](https://discourse.slicer.org/t/ukf-tractography-vs-mrtrix-csd-in-my-data/33165/3 "2023-12-05T00:55:12Z")

</div>

Thank you. I found that SlicerDMRI can’t handle GE scan headers well?My data has 0,1000,2000 b-values, but I found that using DicomToFSL in DWIConvert, I only get 0 and 2000 b-values.

---

<div class="post-metadata">

**Author:** ![pieper](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/pieper/32/8_2.png) [@pieper](https://discourse.slicer.org/u/pieper)\
**Post date:** [December 5, 2023, 7:10pm UTC](https://discourse.slicer.org/t/ukf-tractography-vs-mrtrix-csd-in-my-data/33165/4 "2023-12-05T19:10:12Z")

</div>

Diffusion scans are notoriously tricky. We suggest [dcm2niix](https://github.com/rordenlab/dcm2niix) since it is actively maintained and DWIConvert is very old at this point.

---

<div class="post-metadata">

**Author:** ![gaunny](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/gaunny/32/67838_2.png) [@gaunny](https://discourse.slicer.org/u/gaunny)\
**Post date:** [December 6, 2023, 9:18am UTC](https://discourse.slicer.org/t/ukf-tractography-vs-mrtrix-csd-in-my-data/33165/5 "2023-12-06T09:18:55Z")

</div>

Thank you and it works well now.