ADA4MIA: Active Domain Adaptation for Medical Image Analysis

• ADA4MIA is a benchmark repo dedicated to enhancing Domain Adaptation and Active Learning in medical image analysis. Our goal is to foster robust model development across varied medical datasets, facilitating straightforward evaluation and comparison of different methods.

• This repository collects various state-of-the-art methods, open-source code, and related datasets for the community. If you are interested, you can push your implementations or ideas to this repo or contact me ([email protected]) at any time.

• This project was originally developed for our previous works. Now and future, we are still working on extending it to be more user-friendly and support more approaches to further boost and ease this topic research. The parts of the code for training the source model, generating pseudo-labels, and fine-tuning the target model are provided in our [STDR] project. If you use this codebase in your research, please cite the following works:

     @article{wang2024dual,
      title={Dual-reference source-free active domain adaptation for nasopharyngeal carcinoma tumor segmentation across multiple hospitals},
      author={Wang, Hongqiu and Chen, Jian and Zhang, Shichen and He, Yuan and Xu, Jinfeng and Wu, Mengwan and He, Jinlan and Liao, Wenjun and Luo, Xiangde},
      journal={IEEE Transactions on Medical Imaging},
      year={2024},
      publisher={IEEE}
    }
  
    @article{wang2024advancing,
      title={Advancing UWF-SLO Vessel Segmentation with Source-Free Active Domain Adaptation and a Novel Multi-Center Dataset},
      author={Wang, Hongqiu and Luo, Xiangde and Chen, Wu and Tang, Qingqing and Xin, Mei and Wang, Qiong and Zhu, Lei},
      journal={arXiv preprint arXiv:2406.13645},
      year={2024}
    }

Outline

• [ADA4MIA: Active Domain Adaptation for Medical Image Analysis]
  • 1. Datasets
  • 2. Source Free Active Domain Adaptation
  • 3. Traditional Domain Adaptation
  • 4. Active Learning

1. Datasets

Short name	Paper	Source	Data Link
STDR	Dual-Reference Source-Free Active Domain Adaptation for Nasopharyngeal Carcinoma Tumor Segmentation across Multiple Hospitals	TMI 2024	[dataset]

2. Source-Free-Active-Domain-Adaptation

Short name	Paper	Source	Code Link
STDR	Dual-Reference Source-Free Active Domain Adaptation for Nasopharyngeal Carcinoma Tumor Segmentation across Multiple Hospitals	TMI 2024	[code]
CUP	Advancing UWF-SLO Vessel Segmentation with Source-Free Active Domain Adaptation and a Novel Multi-Center Dataset	MICCAI 2024	[code]
UGTST	An Uncertainty-guided Tiered Self-training Framework for Active Source-free Domain Adaptation in Prostate Segmentation	MICCAI 2024	[code]

3. Traditional Domain Adaptation

Short name	Paper	Source	Code Link
AdaptSeg	Learning to adapt structured output space for semantic segmentation	CVPR 2018	[code]
AdvEnt	Advent: Adversarial entropy minimization for domain adaptation in semantic segmentation	CVPR 2019	[code]
UncertainDA	Uncertainty reduction for model adaptation in semantic segmentation	CVPR 2021	[code]
Tent	Tent: Fully test-time adaptation by entropy minimization	ICLR 2021	[code]
DPL	Source-free domain adaptive fundus image segmentation with denoised pseudo-labeling	MICCAI 2021	[code]
FSM	Source free domain adaptation for medical image segmentation with fourier style mining	MIA 2022	[code]
AdaMI	Source-free domain adaptation for image segmentation	MIA 2022	[code]
CBMT	Source-free domain adaptive fundus image segmentation with class-balanced mean teacher	MICCAI 2023	[code]
CPR	Context-Aware Pseudo-label Refinement for Source-Free Domain Adaptive Fundus Image Segmentation	MICCAI 2023	[code]

4. Active Learning

Short name	Paper	Source	Code Link
AADA	Active adversarial domain adaptation	WACV 2020	[code]
MADA	Multi-anchor active domain adaptation for semantic segmentation	ICCV 2021	[code]
MHPL	Mhpl: Minimum happy points learning for active source free domain adaptation	CVPR 2023	[code]
CLAUS	Hybrid active learning via deep clustering for video action detection	CVPR 2023	[code]
ActiveFT	Active finetuning: Exploiting annotation budget in the pretraining-finetuning paradigm	CVPR 2023	[code]
BAL	Bal: Balancing diversity and novelty for active learning	TPAMI 2023	[code]
ALFREDO	ALFREDO: Active Learning with FeatuRe disEntangelement and DOmain adaptation for medical image classification	MIA 2024	[None]