Fiber Traking, DTI

[ramin.azodi]

Hello All,

Actually in the last step and after fusion when I am trying to work with DTI data. I selected the “Global fiber tracking method” to perform structural analysis. After 4 days my Matlab 2014.b (16 GB RAM, Core 5) is still busy, is that normal?. And when I selected “deterministic Fiber traking” I am getting following error

Create the b matrices
Voxel intensity to absorption conversion
Create B matrix vector [Bxx,2*Bxy,2*Bxz,Byy,2*Byz,Bzz]
Calculate Diffusion tensor, eigenvalues, and other parameters of each voxel
Reference to non-existent field 'BackgroundThreshold'.

Error in ea_DTI (line 100)
            if(S0(x,y,z)>parameters.BackgroundThreshold)

Error in ea_ft_deterministictracking_kroon (line 63)
[ADC,FA,VectorF,DifT]=ea_DTI(DTIdata,parametersDTI);

Could someone tell me how I can solve this problem?

Many Thanks,
Ramin

• This topic was modified 5 years, 11 months ago by ramin.azodi.
• This topic was modified 5 years, 11 months ago by ramin.azodi.
• This topic was modified 5 years, 11 months ago by ramin.azodi.
• This topic was modified 5 years, 11 months ago by ramin.azodi.
• This topic was modified 5 years, 11 months ago by ramin.azodi.
• This topic was modified 5 years, 11 months ago by ramin.azodi.
• This topic was modified 5 years, 11 months ago by andreashorn.

[andreashorn]

Hi Ramin,

honestly, it’s quite possible that the deterministic tracker has issues with some datasets, it’s not well tested and I wouldn’t advise to use it at all (will remove it from Lead-DBS at some point). It’s real diffusion tensor imaging (i.e. with the tensor model) and that’s quite outdated.
The Gibbstracker sometimes takes >1 day but should not take 5 days (except it’s a really huge dataset or something similar). Does it still produce fibers at all? There should be a window that opens which should show you progress. If that’s not the case (i.e. no new window popped up), please abort Matlab and paste the error code here (i.e. where Matlab was hung when you aborted).
Also, there should be a third option which is to use DSI studio under the hood (it’s referred to as generalized Q-imaging (by Fang-Cheng Yeh)). It’s an extremely fast and also modern fiber tracking method which should work on all datasets (i.e. single- or multishell). Can definitely recommend that method – but also can recommend the Gibbstracker (which takes ~4-8 hrs. for one subject on a modern computer).

Also, please use the currentmost version of Lead-DBS since we fixed some fibertracking related things in the recent past.

Let me know if that helps!

Best, Andy

[ramin.azodi]

Thanks a lot for your quick response and the comprehensive answer. My main problem solved now when I updated software and used the Q-sampling. But still i have a problem related to visualization.
when I am following these steps to visualize results, I am getting this message
Add objects ---> Open Trac ---> load FTR.mat

Plotting fibers: K>>
What should I do? If I press continue, I am getting this error,

Plotting fibers:     K>> 1

ans =

     10%Cell contents reference from a non-cell array object.

Error in ea_addobj>addfibertract (line 179)
    if size(thisset{fib},1)~=3

Error in ea_addobj (line 23)
            addfibertract([pana,fina],resultfig,addht,fina,[],0,options);
 
Error using keyboard
Error while evaluating Menu Callback0%Cell contents reference from a non-cell array object.

Could you help me how I can visualize the results?

Many thanks,
Ramin

• This reply was modified 5 years, 11 months ago by ramin.azodi.
• This reply was modified 5 years, 11 months ago by ramin.azodi.

[andreashorn]

Hi Ramin,

I fear trying to show the whole connectome in Lead-DBS is not so easy – Matlab is not really built for this and in any case Lead-DBS isn’t at the moment.. However, just to have a good look at the whole fibers I recommend to use trackvis (www.trackvis.org). There is a file called FTR.trk inside your patient’s folder which you can load in trackvis to show fibers.
Within Lead-DBS, the way to go would be to visualize only the fibers traversing through a volume of activated tissue.
This video walkthrough goes through the whole process:

Just instead of selecting a group connectome in the convis viewer in the end, select “Patient specific fiber tracts”.

I hope this all works..
Finally, you can also do stats on fibertracts connecting your VTA with other areas in the brain. I.e. find out whether differences in connectivity code for clinical outcome or similar. You can do some of that stuff directly using lead group – which is, to be honest, very unintuitive.. Sorry about that.. trying to improve everything but lack the time and manpower for now..

Best, Andy

[Author]

Posts