[IJCAI' 22] Uncertainty-Guided Pixel Contrastive Learning for Semi-Supervised Medical Image Segmentation.
Tao Wang, Jianglin Lu, Zhihui Lai, Jiajun Wen and Heng Kong
Please refer to requirements.txt
ACDC dataset can be found in this Link.
Please change the dataset root directory in configs/_datasets/acdc_224.yaml.
| Method | Model | Iterations | Batch Size | Label Size | DsC | Ckpt | Config File |
|---|---|---|---|---|---|---|---|
| UGPCL | UNet-R50 | 6000 | 16 | 8 | 88.11 | Link | configs/comparison_acdc_224_136/ugpcl_unet_r50.yaml |
| Mean Teacher | UNet-R50 | 6000 | 16 | 8 | 85.75 | Link | configs/comparison_acdc_224_136/mt_unet_r50.yaml |
-
Segmentation results:
-
Pixel features (t-SNE):
@inproceedings{ijcai2022-201,
title = {Uncertainty-Guided Pixel Contrastive Learning for Semi-Supervised Medical Image Segmentation},
author = {Wang, Tao and Lu, Jianglin and Lai, Zhihui and Wen, Jiajun and Kong, Heng},
booktitle = {Proceedings of the Thirty-First International Joint Conference on
Artificial Intelligence, {IJCAI-22}},
publisher = {International Joint Conferences on Artificial Intelligence Organization},
editor = {Lud De Raedt},
pages = {1444--1450},
year = {2022},
month = {7},
note = {Main Track},
doi = {10.24963/ijcai.2022/201},
url = {https://doi.org/10.24963/ijcai.2022/201},
}