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 HARDI tractography

HARDI tractography
Page:   1  
UserPost

9:16 am
July 13, 2011


seanfw

Member

posts 10
Print this Post
1

Hi Ruopeng et al.,

 

I have diffusion data with 64 directions and a b-value of 3000. Using DTK/Trackvis and the DTI imaging model I've managed to obtain tracks fine but when I try to reconstruct the tracks using the HARDI/Q-ball setting it fails completely (just a lot of messy short tracks which have nothing to do with real anatomy). Can you help me out? Is there something obvious that I am missing?

Are the DTK steps for tracking HARDI data different from those for tracking DTI style?

 

Many thanks in advance,

Sean

12:28 pm
July 20, 2011


Ruopeng

Admin

posts 406
Print this Post
2

Post edited 4:32 pm – July 20, 2011 by Ruopeng

Do you have (0, 0, 0) specified in your gradient table? If so you need to remove it before running hardi recon. Of course you also need to enter "number of b0's" correctly.

Ruopeng

5:32 am
October 12, 2011


irvin

Singapore

Member

posts 3
Print this Post
3

Hi Ruopeng, Sean, et al,

 

I have the same issue as Sean described, my HARDI recon turns out incorrect fibre tracts. I'm using 7 b0 and 61 directions, b=1000. My gradient table includes 7 '0,0,0' terms corresponding to the b0 images.

 

I tried removing the 7 '0,0,0' terms in the gradient table and entered 7 as the number of b0s, but the recon still resulted in incorrect fibre tracts.

 

My b0 images are interspersed throughout the acquisition, and not at the beginning. Do I also need to manually remove the 7 b0 DICOM files before running HARDI?

 

Many thanks,

 

Irvin

11:21 am
October 12, 2011


Ruopeng

Admin

posts 406
Print this Post
4

Well, your case is even more complicated. Since b0 is needed for HARDI reconstruction, you can not just remove them. They need to be at the top of the image series. So what you can do maybe is first convert the dicom files into a 4D nifi files, or a series of 3D nifti files. And then move the b0 volume to the beginning of the series. In the 4D nifti case, you need to shuffle the pixel data so that b0's are the first 7 volumes.

Ruopeng

6:04 am
October 13, 2011


irvin

Singapore

Member

posts 3
Print this Post
5

Hi Ruopeng,

 

Thanks for the speedy reply.

 

Sorting the data as you advised fixed the problem and I'm able to run q-ball on the data now. Thanks! Laugh
Page:   1