Orientation in crossing fibers

Hi MRtrix group ,

I am going to use MRtrix for local diffusion reconstruction and then tracktography. My question is that how many crossing fibers can be estimated for a given voxel. Is there any option to adjust it when we are doing local reconstruction ?

Best
Behnam

MRtrix’ dwi2fod does not try to recover discrete fiber orientations, but estimates the full white matter fiber ODF, which is a continuous distribution of fibers orientations on the sphere. As such there is no option to set the number of fiber orientations to recover per voxel.

While the ODF estimation itself is continuous, and in our CSD-based tractography algorithms we make use of the entire continuous ODF, in a number of contexts we perform a subsequent segmentation of the ODF to obtain a discrete number of fibres per voxel. The options for this process can be seen in the fod2fixel command. However even here, we most commonly do not specify a maximum number of crossing fibres per voxel, instead applying thresholds that each fibre must survive during the segmentation process.

Not sure whether or not this will be relevant to exactly what you’re trying to achieve; feel free to give more details here if your path forward is still unclear.

Rob

1 Like

Hello Rob,

I am wondering how can we get ODFs of each imaging voxel. Some voxels may have single fiber, whereas some voxels might have double or triple fibers. I would like to get those single, double or triple fiber voxels separately. Would you please elaborate on this topic a little more? How can we use fod2fixel command to solve this issue?

image

Best Regards,
Suren

The fixel format stores the number of fibers as the first volume in index.mif (see Fixel image (directory) format — MRtrix 3.0 documentation for a detaield explanation of the fixel format).

So this should do the trick:

fod2fixel fod.mif fixel
mrconvert -coord 3 0 fixel/index.mif number.mif

number.mif now contains the number of fibers within each voxel.

1 Like

Thank you very much @bjeurissen sir. :smiley: