How to quantify the similarity of two subjects FODs in template space?

Hello all,

I have created a FOD template of my subjects using the MSMT-CSD pipeline.
How can I quantify how similar are two different subjects FOD when I register them into the template space?

Thank you,

1 Like

Hello Mahmoud,

That depends on your definition of similarity. If you are interested in the possibly masked per-volume mean squared difference (or negative cross-correlation for 3D images), you can use mrmetric. If you are interested in the angular power spectrum, you can use sh2power (using the -spectrum option) for instance applied to the difference image. If you have anything else in mind, mrcalc and mrmath are quite flexible tools.


Thanks, Max. Can the difference image being calculated by mrcalc or mrmath? I guess part of my confusion is because it’s not clear to me what is exactly a FOD map. A difference image makes totally sense when dealing with two scalar image (like FA map) but I’m not sure how to think about the FOD map difference? Please advise.


How can I quantify how similar are two different … FOD(s) …

Shameless self-plug:

While this is intended for improved mapping of fixel data to template space, the underlying cost function that’s used in the optimisation can be thought of as “quantifying the difference between two FODs” (or at least the discrete fixel representation thereof).


Is this currently available in mrtrix?

No, unfortunately it’s going to be a while before I can get around to resolving those developments and getting them into a form suitable for wider release.

I’m already in the process of releasing many technical developments into the wild that I’ve not yet published, so right now I need to focus on bringing my manuscripts up to date.

I would however be interested if you think you may have an application where this method would enable interrogating some question that is otherwise inaccessible; feel free to PM me if that’s the case.