Closed ebrahimebrahim closed 5 months ago
After giving it some thought, I think these are the main methods we should evaluate:
This covers the 3 main registration types (scalar, tensor, fod) and includes the most popular packages we would be expected to compare against. After thinking about it, I would hesitate to compare with any previous version of Hasting's research since it will have a very similar formulation to "ours". It also seems a bit redundant since it would be the same flavor as evaluating voxelmorph as both models were not trained explicitly for dwi.
Great list, and I very much agree with why you pared it down to this set of methods!
The inputs we need to prepare in order to run these evaluations are
list pairwise registration methods that we will evaluate