Which FOD file to use for TractSeg in subject space

[Tanja]

Hi everyone,

I have been trying to extract average tract-wise FD, FC, and FDC and so far I’ve been running TractSeg on the template FOD since the FD/FC/FDC maps are in template space too and the tract masks can be applied conveniently.
I would now like to try running TractSeg on each subject’s FOD individually, the rationale being that segmenting individual images would increase accuracy especially in patient brains with pathological changes. I haven’t seen this done in any publication so I’m wondering if this works well at all.
Since the FD maps are in template space I will need to either a) warp the TractSeg outputs to the template space somehow (how?), or b) run TractSeg on individual FODs in template space in the first place. I considered b) using fod_in_template_space_NOT_REORIENTED.mif which is generated in step 13 in the FBA tutorial and which is in template space. However, I’m not sure this is correct since the tutorial mentions at that step “Note that here we warp FOD images into template space without FOD reorientation, as reorientation will be performed in a separate subsequent step (after fixel segmentation)”. I understand this as the FODs being registered to template space correctly but having the original orientation, and that running TractSeg on these files would be wrong.

Does anyone have any additional insights how to do this? And has anyone compared these two options and could weigh in? I’m afraid that any additional specificity gained from using subject level FODs instead of the template FOD will be meaningless if I have to warp the tract masks or the subject FODs to achieve this.

I have seen something similar being discussed here Fixel Based Analyses in Given Tract but it’s not quite the same question.

Thank you in advance!
Tanja

[smeisler]

@Tanja

However, I’m not sure this is correct since the tutorial mentions at that step “Note that here we warp FOD images into template space without FOD reorientation, as reorientation will be performed in a separate subsequent step (after fixel segmentation)”. I understand this as the FODs being registered to template space correctly but having the original orientation, and that running TractSeg on these files would be wrong.

Correct.

I’d run TractSeg as normal in subject space and then warp the streamlines to template space with tcktransform — MRtrix3 3.0 documentation. This is preferred over warping FD/FC/FDC since the only thing you are manipulating are valueless streamlines, so no worry about data interpolation.

Best,
Steven