Region-based training in MONAI?

[jqmcginnis]

Hello,

I recently started using MONAI for different medical imaging tasks, and really enjoy it. Thanks to the community ! 😄

Having worked with region-based training in nn-unet, I was wondering if something similar exists in MONAI and if someone could give me some advice with respect to its implementation. For a description of region-based training, please refer to this link below:

https://github.com/MIC-DKFZ/nnUNet/blob/master/documentation/region_based_training.md

Prior to asking this question, I took a look at the discussion forum and did some digging, however, if I missed something in the API, please let me know!

Thank you very much.

Cheers,

Julian

[wyli]

thanks, if you are referring to the label indices mappings there's a transform to achieve this

MONAI/monai/transforms/utility/array.py

Lines 1201 to 1208 in 2daabf9

	class MapLabelValue:
	"""
	Utility to map label values to another set of values.
	For example, map [3, 2, 1] to [0, 1, 2], [1, 2, 3] -> [0.5, 1.5, 2.5], ["label3", "label2", "label1"] -> [0, 1, 2],
	[3.5, 2.5, 1.5] -> ["label0", "label1", "label2"], etc.
	The label data must be numpy array or array-like data and the output data will be numpy array.
	"""

or a more specific version in the tutorial https://github.com/Project-MONAI/tutorials/blob/main/3d_segmentation/brats_segmentation_3d.ipynb