George Yiasemis

New loss functions, refactored engines to only implement forward_method, quality fixes

1

Implements: * New loss functions (`NMSE`, `NRMSE`, `NMAE`, `SobelGradL1Loss`, `SobelGradL2Loss`) * `mri_models` performs `_do_iteration method`, child engines perform `forward_function` which returns output_image and/or output_kspace * Changes/Additions in `mri_transforms` * Padding...

**Is your feature request related to a problem? Please describe.** No. **Describe the solution you'd like** Have a single `_do_iteration` in `mri_models` and every child class to implement smt like...

Store val metrics in Tensorboard correctly

1

Validation metrics first reduced, then add to Tensorboard if main process

**Describe the bug** Tensorboard and logger (validation) metrics do not match. **To Reproduce** Any validation process. **Expected behavior** Tensorboard and logger (validation) metrics to match.

Refactors DirectEnums to uppercase

1

Refactors attributes of `DirectEnum`-derived classes to uppercase.

Adds vSHARP model and code quality changes

1

- **vSHARP Model**: Supporting both 2D and 3D. (Details: [vSHARP Paper](https://arxiv.org/abs/2309.09954)) - **3D Reconstruction Experiments**: Various additions, such as transforms, engine changes (e.g., visualization), and the addition of new metrics...

Add new model vSHARP & CMR challenge along with 3D reconstruction code

2

- **vSHARP Model**: Supporting both 2D and 3D. (Details: [vSHARP Paper](https://arxiv.org/abs/2309.09954)) - **CMRxRecon Challenge Code**: Includes necessary code and dataset for the CMRxRecon challenge. Learn more about the challenge [here](https://cmrxrecon.github.io/Challenge.html)....