Support iterative registration

[mathpluscode]

Subject of the feature

What are you trying to do and how would you want to do it differently?

Given a pair of moving/fixed images, we want to perform multiple times of registration to enable larger displacements. In other words, once registered, we can register the warped image to the fixed image again.

This is a common setting in inference. @YipengHu what do you think?

[YipengHu]

I would not consider this is common and the benefit isn't clear or convincing to me. What's stopping user to do this with existing code by running twice?

[mathpluscode]

I would not consider this is common and the benefit isn't clear or convincing to me. What's stopping user to do this with existing code by running twice?

Well, you still need to reorganize the input/output files.

The change should be minimal as you only need to call the following again by redefining the inputs

outputs = model.predict(x=inputs, batch_size=batch_size)

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.