hi-fishu / sau-net

SAU-Net

This is the source code for the paper, SAU-Net: A Unified Network for Cell Counting in 2D and 3D Microscopy Images and this paper is an extended version of our prior work SAU-Net: A Universal Deep Network for Cell Counting.

Our 2D U-Net implementation is based on https://github.com/jakeret/tf_unet.

Dependencies

• python 2.7
• tensorflow (1.15.2)

Data

All the five datasets used in the paper are provided for convenience in https://drive.google.com/drive/folders/1Ap91365akA1FkuWLv9k_DHt_EtrFlrbY?usp=sharing

Download the data folder and put it in the root folder, like this:

sau-net
  |-data
  |  |-VGG
  |  |-MBM
  ...

The dot annotations are processed using scipy.ndimage.gaussian_filter.

Original Datasets:

• VGG
• MBM & ADI
• DCC
• MBC

Run

From the root folder, run

bash run.sh [2D_dataset] [SELF_ATTN_FLAG] [GPU_ID] 

or

bash run_3d.sh [3D_dataset] [SELF_ATTN_FLAG] [GPU_ID] 

For example, the following code will run on vgg dataset with Self-attention module using GPU 0 (the default ID If only one GPU is available).

bash run.sh vgg 1 0

Each time the training and test set will be randomly split by a random seed appended in the output folder. The corresponding model weights and the predictions can be found in outputs/. If you run into memory issues, consider using a smaller batch size, which can be found in the scripts, run.sh or run_3d.sh.

If you find this code useful in your research, please cite our paper:

@ARTICLE{9456970,
  author={Guo, Yue and Krupa, Oleh and Stein, Jason and Wu, Guorong and Krishnamurthy, Ashok},
  journal={IEEE/ACM Transactions on Computational Biology and Bioinformatics}, 
  title={SAU-Net: A Unified Network for Cell Counting in 2D and 3D Microscopy Images}, 
  year={2021},
  volume={},
  number={},
  pages={1-1},
  doi={10.1109/TCBB.2021.3089608}}
  
@inproceedings{Guo:2019:SUD:3307339.3342153,
 author = {Guo, Yue and Stein, Jason and Wu, Guorong and Krishnamurthy, Ashok},
 title = {SAU-Net: A Universal Deep Network for Cell Counting},
 booktitle = {Proceedings of the 10th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics},
 series = {BCB '19},
 year = {2019},
 isbn = {978-1-4503-6666-3},
 location = {Niagara Falls, NY, USA},
 pages = {299--306},
 numpages = {8},
 url = {http://doi.acm.org/10.1145/3307339.3342153},
 doi = {10.1145/3307339.3342153},
 acmid = {3342153},
 publisher = {ACM},
 address = {New York, NY, USA},
 keywords = {cell counting, data augmentation, neural networks},
}