I would like to know if we can do Anatomically constrained tractography (ACT) for non-brain tissue i.e. muscles of leg.
I have the T1 scan, segmented muscles and tendons mask and have already coregistered them to the diffusion data. But I don’t know how can I proceed with the ACT such that the boundary of the tendon and muscle can be used as seed/end for fiber tracking in the respective muscle. The -act in the tckgen command requires the 5TT format image file. Can any one guide me how to generate the input 5TT image file incase of non-brain data.
I gave those authors some advice at the time regarding generation of a 5TT image, but don’t know if they ended up writing a command fit for wider distribution.
The format is ultimately pretty simple. The format is described here. With data coming from some other source, it’s likely just a matter of concatenating 3D images in the correct order. I can’t really give more detailed instructions than that for non-brain data generally because it’s so context-specific. I can only recapitulate that it’s not strictly necessary that there be correspondence between the description of the tissues and the descriptions ascribed to the format; it’s just about controlling which priors are applied to the streamlines at each location.
I am one of the authors on the paper you mentioned and I can give further details on how we implemented the ACT framework for fascicle reconstructions in skeletal muscle.
As you mentioned, to use the ACT framework you need to create a 5tt mask, which for brain tractography consists of masks for white matter, gray matter, csf, etc. For muscle DTI we can use muscle and aponeurosis masks instead. In our case, we created a “white matter” mask by subtracting the binary aponeurosis mask (let’s call this image “a”) from the binary muscle mask (“m”): wm = m - a. We actually used the “not” and “and” operators to ensure that the white-matter mask only had positive values.
mrcalc m.mif a.mif -not -and wm.mif
This will create the mask wm.mif in which tracts are allowed to propagate.
As a grey matter mask, we used the inverted white-matter mask: gm = 1 - wm, or:
mrcalc 1 wm.mif -sub gm.mif
This will create the mask gm.mif. Tracts that enter this region will be terminated.
We didn’t use any other masks so we filled the 5tt mask with empty (zero-filled or “zero”) masks of the same dimensions. Here’s one way of creating those:
mrcalc wm.mif 0 -lt zero.mif
We then concatenated them together into a 5tt mask as follows:
mrcat gm.mif zero.mif wm.mif zero.mif zero_img.mif 5tt_mask.mif -datatype float32
You can use 5ttcheck to make sure the mask is setup correctly for ACT.
5ttcheck 5tt_mask.mif
If all is good, you can use the MRtrix tractography commands with the “-act 5tt_mask.mif” option. We used the deterministic algorithm (-algorithm Tensor_Det) with the white-matter mask as a seed region (-seed wm.mif).