Implement registration framework #6
Conversation
|
stitch_reduce -> displacement_field_reduce |
|
dask array argument -> chunk size specification that is used for internal dask array construction |
|
Add a comment noting that the current readers / writers will work with any itk-supported image type, but streaming will require a streaming supported reader, e.g. uncompressed MetaImage or Zarr. |
|
@thewtex Following up from discussion yesterday, documentation on the Dask "threads" scheduler is here: https://docs.dask.org/en/stable/scheduling.html#local-threads Parallelism occurs within a single process. I have not run benchmarks to determine whether we are seeing speedup with ITKElastix between "synchronous" and "threads" scheduling. |
Update `.gitignore` with common files to ignore
Update `itk-dreg` interfaces in `itk_dreg.base` with docstrings and feature improvements following implementation and testing.
Add `itk_dreg.register` submodule implementing Dask pairwise registration scheduling. Rebased after development and testing.
Add the `itk_dreg.elastix` subpackage implementing pairwise subimage registration with ITKElastix. Includes testing.
Adds `itk_dreg.reduce_dfield` subpackage implementing subimage registration result reduction to a displacement field transform. Includes testing.
Adds `itk_dreg` pytests and mock helpers. Rebased following development.
Adds `pyproject.toml` for packaging `itk-dreg` with the tool `flit`.
Adds examples demonstrating how to use `itk-dreg` for distributed massive image registration.
Add CI workflows to automatically run linting and tests on contributions to `itk-dreg`.
Update `itk-dreg` project documentation with general discussion, developer information, and other useful information.
tbirdso
force-pushed
the
registration-impl
branch
from
bfe4fc1 to
175362a
Compare
|
@thewtex Ping: this PR is ready for review and then merge. Additions are significant, I leave it up to you to decide how closely you'd like to do code review following our previous walkthroughs together offline. |
Implements the
itk-dregout-of-memory registration framework. Includes installable modules, testing, and examples.Refer to updated README and CONTRIBUTING documentation along with notebook examples to get started.
Deferred: Deferred to #7: Run distributed with
LocalCluster. Depends on InsightSoftwareConsortium/ITK#4267.cc @thewtex