[Home] [Paper]

• System (tested on Ubuntu 18.04LTS)
• NVIDIA driver 430
• Python>=3.6
• PyTorch>=0.4
• MATLAB

Python setting

conda env create -f=requirement.yml
conda activate pytorch

docker build ./docker
sh run_docker.sh

CVPR 2019 Contest on Mitosis Detection in Phase Contrast Microscopy Image Sequences

To use dataset, prease follow the guideline. Now the dataset line was expired. If you want to use the dataset, please ask the contest organizers directly. https://ieeexplore.ieee.org/abstract/document/9328484?casa_token=XLj19UfXiEwAAAAA:TdwkxaQwKywNwzsnDje3GgSL6960XqGUxNVLLXu2RBpWyb85DTy2f1TEqJJYYa4E9SmVjrfEzg

1. Candidate path image extraction based on the brightness

matlab -nodesktop -nosplash -r "candidate_extractor(dataset_directory, './output/')"

1. Generate ground truth from candidate

python generate_ground_truth.py

1. Train V-Net

python train.py

1. Prediction by V-Net

python predict.py

If you use this code for your research, please cite:

@article{nishimura2020spatial,
  title={Spatial-Temporal Mitosis Detection in Phase-Contrast Microscopy via Likelihood Map Estimation by 3DCNN},
  author={Nishimura, Kazuya and Bise, Ryoma},
  journal={arXiv preprint arXiv:2004.12531},
  year={2020}
}