Dear FSL experts, I have a few questions regarding probtrackx:
1. As far as I understand tracktography is bidirectional. So if LGN is my seed and V1 is my target, whatever results I get do NOT represent the connectivity FROM LGN TO V1 but just BETWEEN them. Am I right?
2. The waytotal output is directly correlated with the number of voxels of the SEED but not of the target. Am I right?
3. If I am right in the previous questions, yes, I would expect the waytotal number to be bigger when V1 is my seed compared to when LGN is the seed (V1 has a lot more voxels). However, if I divide the waytotal by the number of voxels (or opposite), shouldn't I get a similar result if tractography is bidirectional? And the fdt_paths, shouldn't they be similar?
4. I did it in both directions and got very different results, even in the fdt_paths. The connection is a lot stronger when using V1 as a seed. Is this only because V1 has more voxels or because there are more connections going from V1 to LGN and less from LGN to V1 (which would mean is not bidirectional). If it is bidirectional, then which results shall I believe? Ok, yes, the fdt_paths look better to me when V1 is a seed. But in another data set that a colleague is working on they look better when LGN is a seed.
5. The mayer's loop goes over the ventricles. I'm getting also some non-zero voxels under the ventricles when V1 is a seed. Is that noise? Does it make sense?
6. V1 and LGN are in freesurfer space. However, as I understand, internally tractography always runs in diffusion space. So the program uses the input transformation and transforms the masks to diffusion and after tractography is done returns the results in freesurfer space. So, if things are really run in diffusion space. Shouldn't I have similar results if I transform all my masks to diffusion space and run it directly there? Well, that's what I did and results are very very different. Why?
Thanks!
|