Unpaired Multi-Domain Stain Transfer for Kidney Histopathological Images
Authors: Yiyang Lin, Bowei Zeng, Yifeng Wang, Yang Chen, Zijie Fang, Jian Zhang, Xiangyang Ji, Haoqian Wang, Yongbing Zhang1630-1637
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments and Results We have evaluated our proposed model over ANHIR dataset (Borovec et al. 2020). The results show that our method can realize multi-domain stain transfer with high quality. |
| Researcher Affiliation | Academia | 1Tsinghua Shenzhen International Graduate School, Tsinghua University 2Harbin Institute of Technology (Shenzhen) 3School of Electronic and Computer Engineering, Peking University 4Department of Automation, Tsinghua University |
| Pseudocode | Yes | Algorithm 1: The training process of our method. 1: for number of training iterations do 2: Select an original image x, and the corresponding original label is Lo. 3: Select a target label Lt. 4: Stage 1. Updating the discriminator... |
| Open Source Code | Yes | Our code and Supplementary materials are available at https://github.com/linyiyang98/UMDST. |
| Open Datasets | Yes | We have evaluated our proposed model over ANHIR dataset (Borovec et al. 2020). |
| Dataset Splits | No | We use four sets (kidney 1, kidney 2, kidney 3, kidney 4) as the training set, and one set (kidney 5) as the testing set... There are 39764 images in the training set ... and 8387 images in the testing set. |
| Hardware Specification | Yes | Our model is implemented with Python based on Py Torch on a computer with Intel(R) Core(TM) i5-10400 CPU, 16 GB RAM, and one NVidia RTX 3090 GPU. |
| Software Dependencies | No | Our model is implemented with Python based on Py Torch |
| Experiment Setup | Yes | Our model is trained for 300000 iterations. During training, we use the Adam (DP and Ba 2015) optimizer with β1=0.5 and β2=0.999. The learning rate is set as 0.0001 initially and decreases using linear decay after 150000 iterations. Meanwhile, the batch size of the training dataset is set to 1. |