search

brain-microstructure-exploration-tools / abcd-registration-experiments

experiments with diffusion image registration
2 stars 1 forks source link

Create registration evaluation framework #7

Closed jamesfishbaugh closed 2 months ago

jamesfishbaugh commented 3 months ago

Adding functionality towards a framework to evaluate dwi registration methods.

jamesfishbaugh commented 3 months ago

I noticed it saves the affine as source_dti.aff. Does it save the nonlinear warp somehow? We would need that to later transform tracts.

I had thought it saved a displacement field as well representing the final transformation. I have seen such files saved during processing but maybe are deleted after convergence. I'll take a look if there is a parameter I can include to make sure the warp is saved.

I didn't look at the images and see if they were well-aligned or not. Did you look? Did it look reasonable?

I looked at the results and they looked pretty good. The size and shape of the brain were well matched and some internal structures like ventricles looked decent. They weren't in perfect correspondence though. I am using default parameters from the tutorial so there is room to explore.