Elastic Interaction-based Loss

This is the official code for the paper "An Elastic Interaction-Based Loss Function for Medical Image Segmentation" presented at MICCAI 2020 (https://arxiv.org/abs/2007.02663).

Update 2023.07

• Updated the code to Pytorch 1.11.0
• Fixed some bugs and improved performance

Requirements

• Python 3.7 or higher
• Pytorch 1.11.0 or higher

Usage

To train your model, run the following command:

python train.py --train_dataset '/your_training_data_path' --test_dataset '/your_test_data' --save_path '/save_model_path'

You can also specify other arguments such as batch size, learning rate, number of epochs, etc. See train.py for more details.

If you want to customize your Dataset: modify the ImageToImage2D in ./unet/dataset.py.

The elastic interaction loss file is located in ./unet folder. You can import it and use it as a custom loss function for your segmentation model.

Citation

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

@inproceedings{LanXZ20,
  author       = {Yuan Lan and
                  Yang Xiang and
                  Luchan Zhang},
  title        = {An Elastic Interaction-Based Loss Function for Medical Image Segmentation},
  booktitle    = {Medical Image Computing and Computer Assisted Intervention - {MICCAI}
                  2020 - 23rd International Conference, Lima, Peru, October 4-8, 2020,
                  Proceedings, Part {V}},
  series       = {Lecture Notes in Computer Science},
  volume       = {12265},
  pages        = {755--764},
  publisher    = {Springer},
  year         = {2020}
}

Acknowledgements

This code is based on the Pytorch UNet template from https://github.com/cosmic-cortex/pytorch-UNet. We thank the authors for their work.