README.md

GANDLF Augmentation

Adding a new algorithm

• Define a new submodule under GANDLF.data.augmentation.
• If new algorithm is a class, follow principle of adding torchio transforms.
• If new algorithm can be wrapped in a function:
  • Keep in mind that unless otherwise specified, the functional transforms work on images AND segmentations (similar to torchio.transforms.SpatialTransform), thus, if an algorithm is meant to work only on images, add it as a class by inheriting from torchio.transforms.IntensityTransform.
  • Add function in new submodule.
  • Ensure it returns a torch.Tensor.
  • If the new function is from a pre-defined package, put it under GANDLF.data.augmentation.wrap_${package_name}.py.
• Wrap the new algorithm in a _transform function as shown in GANDLF.data.augmentation.crop_external_zero_planes.crop_external_zero_planes_transform. This is the algorithm's identifier for GaNDLF.
• Add the algorithm's identifier to GANDLF.data.augmentation.__init__.global_augs_dict as appropriate.
• Call the new algorithm from the config using the data_augmentation key.