Compatibility of nitransforms and AFNI

[oesteban]

Compatibility of nitransforms and AFNI

• Oscar Esteban (@oesteban), Stanford University
• Stefano Moia (@smoia), BCBL

---

Brief description of what was accomplished with this project

https://github.com/poldracklab/nitransforms contains a prototype to deliver a proof of concept and support the development of the BIDS Derivatives extension about spaces and transforms.

In particular, two issues were planned to be addressed:

• Implement AFNI affines on oblique datasets nipy/nitransforms#45 - Make nitransforms understand and communicate with AFNI when one or two datasets involved in an affine transform are oblique (and hence AFNI "deobliques" them).
• Read in 3dQwap deformation fields nipy/nitransforms#5 - Make nitransforms capable of reading and applying displacements fields generated by AFNI tools.

Results

1. Implement AFNI affines on oblique datasets nipy/nitransforms#45 - With the very much appreciated assistance of @afni-dglen, we are now capable of replicating "(de)obliquing" matrices, which AFNI writes on the NIfTI extension of deobliqued datasets (see question: how affines are applied to oblique datasets afni/afni#116). Although that was great progress, we are still unable to replicate AFNI's behaviors when it comes to applying transforms or write transforms out that will be used with, e.g., 3dAllineate. To follow up on this second problem, Concatenate affine transforms afni/afni#129 was opened and @mrneont is helping me understand how AFNI uses these matrices.

2. Read in 3dQwap deformation fields nipy/nitransforms#5 - @smoia managed to implement the functionality, with the only exception of oblique datasets (which are waiting on the previous issue).

---

Resources

All resources and people are linked above.

---