Weird fibertracking results

45 views
Skip to first unread message

Vesna

unread,
Apr 25, 2012, 4:38:47 PM4/25/12
to cmtk-...@googlegroups.com, Miguel Lechón, cristianubgm
So I have been having the same problem with the cmp with all my DSI and now DTI data. The results of the fibertracking look weird.

If i open the file that is generated by the cmp in the directory CMP/fibers/streamline.trk I get really weird results (see image cmp_tracking.png)

If however I use DTK and do tracking with the default parameter on the file CMP/raw_diffusion/DTI_resampled_2x2x2.nii.gz I get results that look more normal (original.png)

I believe this problem comes from the dti_dir.nii. Can someone help on this? What is the meaning of this file anyway?

I would like to share my data (perhaps with Alessandra) and then to see if the problem persists on your side?I have the same issue when I use our DSI data as well and also some example data. 
I have posted about this before and then the hints were that the problem was in the registration step but I don't think that is the problem. It has been annoying since I cannot generate connectomes for already two months due to problems on various steps on the pipeline. Now that I am getting close I am stuck on this. 

Thanks for your help!!!

Vesna
cmp_tracking.png
original.png

David Grayson

unread,
Apr 25, 2012, 6:18:08 PM4/25/12
to cmtk-...@googlegroups.com
Hi Vesna,

In your cmp image, you're showing only one y slice, but in your 'original', you've shown the whole brain, which doesn't look quite right either. Can you show just a y slice from the 'original' ? You can use the 'slice' tool in trackvis to do this.

It looks to me like either you might be using the wrong gradient table, or you need an inversion flag in the DTK step. It's critical that you know the gradient table you are using is accurate. You could also try adding '--ix' to your DTB_dtk2dir parameter

-David

2012/4/25 Vesna <vesna.p...@gmail.com>

Qingyang Li

unread,
Apr 25, 2012, 11:46:43 PM4/25/12
to cmtk-...@googlegroups.com
Hi Vesna,

You already got the fibers. The reason it looks odd is that you have your "slice filter" on. Turn it off then you will be able to see the whole brain fibers...

Yang

Alessandro Daducci

unread,
Apr 26, 2012, 2:20:32 AM4/26/12
to cmtk-...@googlegroups.com
Hi,

can you share a dataset so that we can check? We've recently found a bug (not on our side) in the way dtk stores the results; they changed the voxelsize ordering in the header passing from version 0.6 to 0.6.1 (so, late november 2011). Can you check which version you are using?

Alessandro

Vesna

unread,
Apr 26, 2012, 5:46:59 AM4/26/12
to cmtk-...@googlegroups.com
Hi Alessandro,

With who to share the data? Can i send it to you and Alessandra via wetransfer for example?

I tried to correct the gradient table , in trackvis the results look reasonable but with the cmp again really weird...

Tnx,
Vesna

Alessandro Daducci

unread,
Apr 26, 2012, 9:16:01 AM4/26/12
to cmtk-...@googlegroups.com
hi Vesna,

Can you share the data with the rwo of us? Unfortunately I'm travelling right now and I can have a look as soon as I arrive home, but perhaps Alesssndra can already have a look.

Btw, both umof us will be in Barcelona, so we can discuss there as well if you need.

Cheets,
Alesssndro

Reply all
Reply to author
Forward
0 new messages