Hi @synchon! Thanks for your interest. The pairwise transformations have been part of the vision for templateflow, but have fallen by the wayside with nobody really making it a priority to get them done.

We have some code for performing the registrations (undoubtedly out-of-date and needing some updating): https://github.com/templateflow/registration-framework/

My main concern is that I don't know that we have a solid QC mechanism to validate the registrations. The current registration framework uses ANTs SyN registration, though MMORF might be worth a look.

Anyway, I think overall we're interested in including this, and I would be interested in your thoughts on QC and registration frameworks.

This would indeed be a very desirable issue to solve. Let me know if somebody is willing to work on this - over the next year we need to look at this in more detail anyway, so I'd love to discuss how to solve this with high quality.

Joern

Hi @effigies ! Apologies for the delayed response.

My main concern is that I don't know that we have a solid QC mechanism to validate the registrations.

That is indeed a problem we have as well and for now have relied on manual validations to make sure the registrations are correct. You can find our current transformation files here: https://gin.g-node.org/juaml/human-template-xfms .

The current registration framework uses ANTs SyN registration, though MMORF might be worth a look.

We've also relied on ANTs SyN registration similar to https://github.com/templateflow/registration-framework/ . MMORF does look very interesting, thanks for sharing! In general, I find registrations by ANTs good enough given but have to admit that I'm not an expert in it.

Anyway, I think overall we're interested in including this, and I would be interested in your thoughts on QC and registration frameworks.

We've been planning to validate the transformation files using https://github.com/afids/afids-data/tree/main and I'm curious what you think about it. Apart from this, my knowledge of QC is limited, but really interested to know if there are better alternatives.