50 dataset adaptor to make oasis unet run on fast mri dataset

[MatthiasLen]

@ydu0117 as discussed: the dataset adaptor to allow OASIS model execution on FastMRI

• i only tested on sigle coil knee: as expeced the quality is mixed because fast MRI kneee is out of domain for OASIS Unet
• related issue Add multicoil MRI #23 (multi-coil data support via DeepInverse). I have tagged @Andrewwango there . Maybe now that the adaptor is there, it is easy to close this out so that FastMRI Brain can be processed by OASIS

[Copilot]

Pull request overview

This pull request adds an adapter and inference selection layer so the packaged OASIS single-coil U-Net can be run on FastMRI measurements by converting FastMRI k-space into OASIS’s centered FFT convention, and updates the example CLI and tests accordingly.

Changes:

• Introduces FastMRI→OASIS k-space conversion helpers and exposes them via mri_recon.utils.
• Adds a centralized reconstructor-selection module (mri_recon.reconstruction.inference) with explicit algorithm names and dataset/algorithm compatibility checks.
• Refactors the FastMRI inference plotting example to support explicit U-Net variants and the centered-OASIS path on FastMRI.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
tests/test_utils_io.py	Adds tests verifying FastMRI→image and FastMRI→OASIS-kspace helpers match the native adjoint/centered path.
tests/test_reconstructions.py	Switches tests to use choose_reconstructor/compatibility helpers and adds coverage for algorithm selection logic.
README.md	Updates CLI examples to use explicit algorithm names (e.g., unet-fastmri, unet-oasis-acceleration8).
mri_recon/utils/oasis_adapter.py	Adds conversion helpers to map FastMRI measurements into the OASIS centered FFT convention.
mri_recon/utils/__init__.py	Re-exports the new adapter helpers as part of the public utils API.
mri_recon/reconstruction/inference.py	New module defining explicit algorithm IDs, compatibility validation, and reconstructor selection.
mri_recon/reconstruction/__init__.py	Exposes the new inference utilities at the package level.
examples/plot.py	New shared plotting helper module for k-space visualization.
examples/fastmri_inference_plot.py	Refactors the example to use explicit algorithm selection and to adapt FastMRI measurements for OASIS-centered inference.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.