You must be logged in to post Login Register


Lost Your Password?
Search Forums:


 






Wildcard Usage:
*    matches any number of characters
%    matches exactly one character

Forum Diffusion Toolkit Usage The best parameters to make a track

The best parameters to make a track
Page:   1  
UserPost

9:03 am
June 10, 2010


Patricia

France

New Member

posts 2
Print this Post
1

Hi,

 

I'm naive in DTI and I have problems to make a track.
I obtain only broken tracts.
I have read that it could be due to the mask and the angle treshold.
I really want to know how could I choose "the orientation patch" and "the propagation algorithm" too.
Can you give me your advice to choose the best parameters???
A range???
And (or) a page with the information above the considerations to make the decision???
I hope you can help me.
Thanks.

 

Patricia

 

12:50 pm
June 10, 2010


Ruopeng

Admin

posts 114
Print this Post
2

Post edited 12:50 pm – June 10, 2010 by Ruopeng

Hi Patricia,

There is no best parameters to make a track.Smile The "best" parameters varies a lot depending on the quality of your data as well as the structures/fiber bundles you're looking for.

When you get tracks that do not look right, it is usually because:

* Masking threshold was set too high. You will see broken tracks and lots of empty spaces in the brain with no tracks inside. You need to load the map in an image viewer and find the proper value for the threshold to mask out background.

* The gradient table you selected was wrong. Usually you get totally non-sense short random tracks when this happens.

* It needs "orientation patch". This usually shouldn't happen if your raw data is from Siemens scanners. For GE data, a lot of times you need to apply an inversion flag ("Invert Y" or "Invert Z"). Unfortunately there is no definitive way to decide what flags to apply automatically. You'll just have to try each of them until the tracks make sense.

Best,

Ruopeng

6:10 am
June 14, 2010


Patricia

France

New Member

posts 2
Print this Post
3

Mr. Ruopeng

Thanks for your observations.

I've tried to do the track with your instructions, but I still have the same problem.

The table of gradients is the correct one and I've tried with different treshold for the mask and the angle, but nothing has worked.

It could have some problem with the fomat of my data?

For to do the track I use Nifti data.

The table of gradients works for Nifti and DICOM?

There is some other thing that can be causing the problem?

Cry

Thanks for your time.

 

Best,

Patricia

 

9:00 pm
June 14, 2010


Ruopeng

Admin

posts 114
Print this Post
4

Would it be possible for you to post a screenshot of the wrong tracks?

1:53 pm
August 6, 2010


silviana

New Member

posts 2
Print this Post
5

Hello !

I just wonder how you solve your problem, I have the same one.. getting broken tracks I have tried to change all parameters, but still the same problem…  any suggestion  ?

 

cheers !!!

 

Silvia Smile

11:02 am
August 10, 2010


Ruopeng

Admin

posts 114
Print this Post
6

Would it be possible to post a screenshot of the wrong tracks? and/or the log text of Diffusion Toolkit?

Ruopeng

8:32 pm
August 12, 2010


silviana

New Member

posts 2
Print this Post
7

Hi Ruopeng,

 

This is a screenshot of the tracks, it is supposed to be the corpus callosum.. and the text of DToolkit.

 

Thank you !!

——— Launching dti_tracker —————–

 

Volume dimension: 128 128 40

Voxel size: 1.797 1.797 3.000

 

Threshold for mask one: 27.1211 451

 

 

Reading tensor data…done.

 

Tracking… 100%

Number of tracks: 126521

Time taken: 00:05:31

 

——— Launching trackvis —————–

 

TrackVis successfully launched.

 

Total time taken: 00:05:35

 

non sense tracks..

2:26 pm
August 13, 2010


Ruopeng

Admin

posts 114
Print this Post
8

Do you have the log text for the reconstruction step? This appears the reconstruction was totally wrong, likely because of wrong gradient table or wrong number of b0's provided.

Ruopeng
Page:   1