Refactor PET-related classes

[jhlegarreta]

What would you like to see added in this software?

PR #174 finished the implementation of PET-related data, model and estimator classes. The implementations deviated from the base class implementations or did not inherit base class implementations:

• The data.PET class implements a lofo_split method. However, there is a lovo_split function in the splitting module. The data splitting function used by PET should dwell in splitting and the necessary refactoring/reconciling should take place so that it is sufficiently generic so as to be used by any modality.
• The data.PET.to_nifti function deviates from the base class signature, and thus, when adding type hinting was attempted in PR STY: Add type hints to PET-related classes and functions #189, mypy complained about this. So the signature should be reconciled with the one from the base class or the latter should be adapted.
• The statement

  nifreeze/src/nifreeze/data/pet.py

  Line 133 in d2def7a

  root = in_file["/0"]

  seems to assume some HDF5 file structure that may not be common across modalities. This should be harmonized.
• The PETEstimator class generates the train/test splits internally. In general, handling the train/test splits should be done by some manager class that is common to all modalities, and thus, should be taken out of the PETEstimator.run function:

  nifreeze/src/nifreeze/estimator.py

  Line 252 in d2def7a

  train_dataset = PET(

• The correction method should probably be made configurable, and default to/offer only the B-Spline method for now. Also, the method itself should probably be refactored into a separate function/class/module.

Do you have any interest in helping implement the feature?

Yes

Additional information / screenshots

No response

[jhlegarreta]

First and last but one items should be addressed by PR #203, and last item should address #204.