Currently there are 2 dataset specific execution paths.
The goal is to be able to consistently evaluate all recon methods on Fast MRI Maybe we can later tackle building a dataset adaptor such that the fast MRI unet can be executed on OASIS.
Idea for a possible solution:
Currently there are 2 dataset specific execution paths.
The goal is to be able to consistently evaluate all recon methods on Fast MRI Maybe we can later tackle building a dataset adaptor such that the fast MRI unet can be executed on OASIS.
Idea for a possible solution:
