ENH dwiregister

[Lestropie]

The issue of inter-modal registration between DWI and T1-weighted image data is a persistent one. It comes up on the forum all the time, and different people have different preferences. There is however a finite set of possible approaches. It would be convenient to have a script utilising the algorithm interface that offers multiple different approaches.

Each would serve the basic purpose of transforming the DWI to align with the T1-weighted image (or potentially other anatomical reference in some applications), rotating the gradient table when it does so. But the tools used to do the registration internally within each algorithm may vary considerably.

Possible algorithms:

• Basic flirt, using mean b=0 image by default, followed by transformconvert
• epi_reg, maybe by default using fsl_anat to derive the brain-extracted T1-weighted image
• The technique I use in MRtrix3_connectome based on the INVERSE method (generate pseudo-T1w from DWI mean b=0 using histogram matching and register to T1w; then same in other direction, ie. generate pseudo-b=0 image based on T1w and register to DWI mean b=0; average the two transforms)
• Generic wrapper for ANTs (I'm not personally familiar with the types of calls people may use here)
• Generation of predicted T1w image using multi-tissue decomposition, register, optimise T1w intensities assigned to tissues, repeat

Given that many users have their own approaches, and may indeed have bundled up such approaches in scripts, once the base command is in place it would be an opportunity for users to contribute their own algorithms for this task to the software.

[Lestropie]

Possible alternative name: dwi2anat? Maybe a different usage of "2" in command names compared to others. Or something longer like dwianatreg.