According to the help, it’s said, - FACT: the directions file (each triplet of volumes is the X,Y,Z direction of a fibre population). I I am confused is it the grad encoding files or anything else? and how can I get it if it’s not the grad encoding file?
Thank you in advance.
Hi @Meng ,
I haven’t used FACT lately, but I suspect it needs the output of
tensor2metric with the
-vec option: that’s basically the field of principal eigenvectors from the tensor. Since FACT doesn’t need any other property from the tensor, it’s (computationally) beneficial to compute those eigenvectors up front. Does that make sense?
What this description means (with a little more verbosity) is that:
- Your input image should have a number of volumes that is a multiple of 3 (i.e. 3, 6, 9, …)
- These volumes are grouped in sets of 3. So, for instance, if the image contains 6 volumes, this is 2 sets of 3 volumes.
- In a particular voxel, for a particular set of 3 volumes, the three image intensities correspond to a direction: the three image values represent the X, Y and Z components of that direction.
It’s possible to have less directions in some voxels than others; e.g. in one voxel you may have the values [1,0,0] (representing a fibre oriented in the X direction), then [NaN,NaN,NaN] (indicating that there are no more fibre directions in this voxel).
Importantly, the representation of these directions has nothing to do with the diffusion gradient encoding. Both are a mechanism for specifying directions, but in this case the directions are different for each voxel, hence why they are represented using image values instead.
MRtrix commands that produce such data include