I would like to perform tractography in a lesioned brain and find out how many fibers between 2 ROIs were directly affected by the lesion. The lesions are mainly WM lesions, however, sometimes they also affect the GM. What would be the best way to perform tractography in this case? I have 3 ideas on how to achieve that:
- Insert the dilated lesion mask in the segmented 5tt T1 image and use this combined image for the fiber tracking. Am I right, that like this, the algorithm would simply ignore the tissue that is lying within the lesion mask? It seems like there is a big disadvantage of this approach: if the number of fibers is set to e.g. 10M, fibers will be “squeezed” in the unlesioned brain, and in case of a very big lesion, this would lead to much more fibers in the unlesioned brain in order to reach the 10M, than if there would not be a lesion. Also, with this approach it might be difficult to track, how many fibers are actually affected by the lesion, as we do not have a comparison to how many fibers were present in the brain before the lesion occurred.
- Do the fiber tracking in the unlesioned healthy hemisphere only and then mirror the tractogram to the lesioned hemisphere to create a “healthy” tractogram, then insert the lesion mask to see how many fibers are affected by the lesion. Like this we simulate how the healthy brain could have looked before the occurrence of the lesion. A disadvantage would be that the tractogram of left and right hemisphere would be identical, which in reality is not the case. However, with this approach we would have a simulated tractogram of how the brain was before the lesion occurred and we could see how many fibers were directly affected by the lesion.
- Run 5ttgen on the unmasked T1 image and correct results manually to make sure the lesioned area is considered as the correct tissue type (mostly WM, and in case of GM lesions, GM). How would the tractography algorithm treat that lesioned area? Would I then insert the lesion mask after tracking the fibers to see how many fibers were affected?
I would really appreciate your input on those ideas to find out the best way to go.
Thank you very much in advance,