Modify Parallel Diffusivity in a voxel-by-voxel basis for stroke modeling

[Ande Bagdasarian]

Is it possible to set the intrinsic parallel diffusivity to not be constant? I am trying to use it in a stroke model, but as the intrinsic parallel diffusivity changes, fixing it in the standard NODDI toolbox will give inaccurate results.  There is a portion of the code in the MakeModel section:

 

diIdx = GetParameterIndex(modelname, 'di');

model.GS.fixed(diIdx) = 1;

model.GD.fixed(diIdx) = 1;

 

I assumed that setting these equal to 0 instead of 1 would do the trick, but it severely underestimated NDI and ODI and zeroes out values in the stroke region I am looking at. My goal is to use a dataset of parallel diffusivities that I calculated previously from the same DTI data, and calculating the NDI and ODI in a voxel-wise basis with the changing parallel diffusivity. Is this possible to do in the existing code, and if not how would you recommend changing it? 

[Hui Zhang]

Hi Ande,

Your change is correctly done.  The issue is that typical multi-shell diffusion data is not rich enough to support the fitting of all the parameters without degeneracy.  You can easily change the fixed diffusivity value to something more appropriate for your data.  A number of postings have been written to explain how this is done.

Gary

--
You received this message because you are subscribed to the Google Groups "NODDI" group.
To unsubscribe from this group and stop receiving emails from it, send an email to noddi+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/noddi/5ecfc769-ace0-4fef-afc6-429489ae80d4%40googlegroups.com.

[Ande Bagdasarian]

Hi Gary,

Thanks for the quick response! 

I'm not quite sure I understand. The existing value that I've been using is fine for naive or non-stroke data that I've processed. I didn't want to just change the fixed diffusivity to a value specifically for the stroke voxels, because then I'd have reprocess again for each stroke dataset to evaluate stroke and healthy tissue, doubling the processing time.

Is there no way to alter the code to bring in a pre-determined dataset of parallel diffusivities for each voxel that has already been calculated to take the place of the diffusivity constant? It would of course have the same number of datapoints and geometry, and just be a matrix of constants that it could iterate through when performing the fit, where it could go through each voxel as that voxel's "constant"? This way the code could have slightly more accurate measures in areas with stroke lesions, and the original code wouldn't have to solve for these values itself, avoiding the degeneracy issue you mentioned. Let me know what you think.

Ande

[Hui Zhang]

You will have to write some code to select the right fixed diffusivity value per voxel and then fit that voxel.

Gary

--

You received this message because you are subscribed to the Google Groups "NODDI" group.
To unsubscribe from this group and stop receiving emails from it, send an email to noddi+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/noddi/92afbd91-1dc1-4da9-988d-4a64e88fa535%40googlegroups.com.