Using Deep Learning for Predicting the Dynamic Evolution of Breast Cancer Migration (Wound Healing Assays Prediction)

This is the official implementation code of the paper "Using Deep Learning for Predicting the Dynamic Evolution of Breast Cancer Migration" (Paper)

[Paper] [Dataset] [BibTeX]

Model architectures

Quantifying Wound Progress:

Predicting Next Frame:

Download last release

Support us downloading our last release

Click on this button to download
We also appreciate your support downloading it on ZENODO site:

Installation

The code requires python>=3.9, as well as pytorch>=1.7 and torchvision>=0.8. Please follow the instructions here to install both PyTorch and TorchVision dependencies. Installing both PyTorch and TorchVision with CUDA support is strongly recommended.

Install wound-healing-PWPF:

pip install git+https://github.com/frangam/wound-healing.git

or clone the repository locally and install with

git clone git@github.com:frangam/wound-healing.git
cd segment-anything; pip install -e .

Dataset

You can download our MCF-7 Dataset at:

Model Checkpoints

Click the links below to download the checkpoint for the corresponding model type.

These models can be instantiated by running:

For predicting next frame:

from tensorflow.keras.models import load_model
from woundhealing.model import ssim, mse, psnr

next_frame_model = load_model("<path/to/checkpoint>", custom_objects={'ssim': ssim, 'mse': mse, 'psnr': psnr)

#you need to provide your frames in gray scale (1, frames, height, width, 1)
new_prediction = next_frame_model.predict(np.expand_dims(frames, axis=0))

For quantifying wound progress:

from tensorflow.keras.models import load_model

quantifying_model = load_model("<path/to/checkpoint>")

#you need to load your X_test and images_tests
predictions = quantifying_model.predict([X_test, images_test])

Wound Segmentation

We use a combination of Meta's Segment Anything model and MedSAM model

Citation

If you use our code in your research, please use the following BibTeX entry:

@article{Garcia-Moreno-PWPF,
  title={Using Deep Learning for Predicting the Dynamic Evolution of Breast Cancer Migration},
  author={Garcia-Moreno, Francisco Manuel and Ruiz-Espigares, Jesus and Marchal, Juan Antonio and Gutierrez-Naranjo, Miguel Angel},
  year={2024},
  journal={Computers in Biology and Medicine},
  doi={10.1016/j.compbiomed.2024.108890},
  note={\url{https://authors.elsevier.com/tracking/article/details.do?aid=108890&jid=CBM&surname=Garcia-Moreno}}
}

And also cite our MCF-7 Dataset used to train our software:

@misc{WHiM-BC_Dataset,
  title={WHiM-BC Dataset: Evolution of CSCs and non CSCs MCF-7 Migration in Wound Healing Assay},
  author={Garcia-Moreno, Francisco Manuel and Ruiz-Espigares, Jesus and Marchal, Juan Antonio and Gutiérrez-Naranjo, Miguel Ángel},
  year={2023},
  doi={10.5281/zenodo.8131123},
  url={https://doi.org/10.5281/zenodo.8131123},
  note = {version 1.0}
}