Filtering Tracks by Curvature

Dear MRtrix Community,

I have done a whole-brain-tracking of my data, but since I’m only interested in the subcortical U-fibres, I’m looking for a way to filter these without using an anatomical mask. Is there any tool to filter tracks for instance by their curvature? Or does anyone have a better idea how I could achieve this?


Welcome Chris!

I happen to have a similar interest at the current point in time, so have myself been thinking about how best to implement such functionality.

There’s currently a feature gap, in that:

  • tckedit does not have the ability to directly calculate mean streamline curvature (or comparable measure) and then select streamlines based on such a measure;

  • tckmap can calculate streamline mean curvature, but only in the context of mapping to a TWI;

  • Even if you could perform some calculation elsewhere to classify streamlines based on whether you do or do not want to select them, tckedit does not have any mechanism by which you could provide such pre-determined information.

I’ve made some progress in altering the software capabilities and interface in order to broaden the scope of possibilities here, but it’s still in private development. So if anybody has any ideas or strong opinions on what they’d like to see in this domain, let me know. I’d also suggest looking at this manuscript; I’ve been thinking about integrating that calculation or something comparable.