Clarifying the parameter "seed_direction" when using iFOD2


Dear MRtrix developers,

I am currently using tckgen to do tractography and I am interested in starting the tracking on a specific direction. This is helpful since in the region that I am tracking different bundles cross, and I am interested in only one of them. For what I saw in the source code, the initial direction that I give is being “rotated”. The code in particular is: “rotate_direction (d, random_direction (max_angle, sin_max_angle));” where the parameters come from here.

To clarify, is this trying to rotate my initial direction in order to bring it closer to a peak in the ODF? If this is the case, I will be very glad actually, since my initial direction is somehow close to the “real one” of the bundle, but is not necessarily very precise.

Thanks for your time!
Best Regards,
Guillermo Gallardo.


Yes and no. This is where the seed direction is set, and it basically sets the direction at the seed point to what you specified. But this is a probabilistic algorithm, and using default settings it’ll sample with a 45° cone about that direction – this is where that tolerance to the initial direction comes from.

Hope that makes sense…?