PyMIC-v0.6.1

Add support of more self-supervised and weakly supervised learning methods.

PyMIC-v0.4.0

Update notes:

1, pymic_net_run train and pymic_net_run test have been replaced with pymic_train and pymic_test respectively.

2, No need to use pymic_ssl, pymic_nll and pymic_wsl, and just use pymic_train to deal with all of these cases.

3, Allow printing version via pymic.version

PyMIC-v0.3.1

Update from previous release:
1, Allow automatic installation of dependencies when using "pip install"
2, For evaluation of segmentation results, support setting a list of evaluation metrics. e.g., "metric = dice" is now updated as "metric_list = [dice, assd]" in the configuration file.

PYMIC-v0.3.0

update algorithms for annotation-efficient learning

PYMIC-v0.2.5

New support for

• Semi-supervised learning
• Weakly supervised learning
• Learning with noisy labels
• Customized inference function

PYMIC-v0.2.4

new features compared with 0.2.3:
1, Support multiple checkpoint ensemble
2, Simplified configure files by allowing default values
3, Update loss functions

PyMIC-v0.2.3

1, Allow a combination of several loss functions for segmentation
2, Set transform and batch size for validation and testing

PyMIC-v0.2.2

Several updates from the last release:

1. support multi-GPU training
2. simplify the command for training and testing
3. enable selecting the latest checkpoint or the best performing checkpoint
4. add ROC for classification taks

PyMIC-v0.2.1: Merge pull request #9 from HiLab-git/dev_cls

Version v0.2.1. New properties: add support to image classification.

PyMIC-v0.1

The first version supporting 2D and 3D image segmentation.