Hi Celine - The path.pd.nii.gz file contains the full probability distribution without any thresholding. This distribution, as with any probabilistic tractography method, expresses the amount of uncertainty in which is the path of most coherent diffusion between the two end regions. (In tracula's case, it's actually both most coherent diffusion and most consistent with the anatomy of the CST, or whichever pathway, based on the atlas.) If in a patient there's widespread degeneration over the entire path, this could manifest itself as more uncertainty and hence a more spread out distribution. So depending on the variability among your subjects you could get wider or narrower pathway ROIs. If you find that a threshold based on path.pd.nii.gz alone is not consistent for your purposes, you could try combining it with an FA threshold.
Hope this helps, a.y
On Wed, 4 Sep 2013, celine@nmr.mgh.harvard.edu wrote:
Dear experts, I read some previous answers about the thresholding of the paths obtained after tracula process, but I am not completely sure I had the right answer. I will need to get an roi from a path for example the cortico-spinal tract that will be consistent amongst patients. I will further use this roi to calculate the lesion volume (my subjects are patients with MS) within a specific path. My question is: is the path.pd.nii.gz file already threshold or should I threshold it to a specific value? If I use the same threshold for each subject for example at 10% of the maximum value, will I obtain an ROI that is consistent among subjects or is is possible that I would get a very thin CST in some subjects, or a very wide cst for example? Thanks for your help Celine