STIFMaps 0.0.6

STIFMaps predicts the local stiffness of breast tissue using the morphology of collagen

Spatially Transformed Inferential Force Maps (STIFMaps)

Contents

-Overview
-Repo Contents
-System Requirements
-Installation Guide
-Reproducing Manuscript Results
-License
-Contact

Overview

Code from the manuscript: "STIFMap employs a convolutional neural network to reveal spatial mechanical heterogeneity and tension-dependent activation of an epithelial to mesenchymal transition within human breast cancers"

Repo Contents

-env.yml: yml file for creating the conda environment to run the STIFMaps code out of a Jupyter notebook
-network_training.ipynb: Jupyter notebook for training the model presented in the manuscript
-stiffnesses.csv: The table of positional stiffness values used for model training
-STIFMaps.ipynb: Jupyter notebook for creating STIFMaps out of an input image using trained networks

Additional data is available via https://data.mendeley.com/datasets/vw2bb5jy99/2
-raw_squares: The images used for training the neural networks
-trained_models: The completed, trained models for use with STIFMaps.ipynb to predict elasticity values on an unknown tissue
-output: Statistics regarding the training and accuracy for the trained models

System Requirements

'STIFMaps' should run on any standard computer capable of running Jupyter and PyTorch, though 16 GB of RAM is required to enable CUDA optimization. Note that the computer must have enough RAM to support in-memory operations and the extent of memory usage depends on the size of the image that the user is trying to characterize using 'STIFMaps'. Within 'STIFMaps.ipynb', the user may downsample the image prior to stiffness predictions to reduce memory consumption.

Typical install time to create the conda environment is about 15 minutes. The runtime to reproduce the network training should take several hours. These runtimes were generated using a linux computer with 64 GB of RAM, 16 cores@3.60 GHz, and running Ubuntu 18.04.

Installation Guide

To generate STIFMaps, create a virtual environment using the 'env.yml' file. Then, refer to the Jupyter notebook 'STIFMaps.ipynb' to create STIFMaps using paired GFP and DAPI images. Additional staining images may optionally be included to evaluate the overlap between stain intensity and STIFMaps.

Reproducing Manuscript Results

To reproduce manuscript results, the Jupyter notebook used for building the neural networks is available via 'network_training.ipynb'. As inputs, the Jupyter notebook needs the elasticity values contained in 'stiffnesses.csv' as well as the image files available at data.mendeley.com with reserved DOI: 10.17632/vw2bb5jy99.2

25 completed models are available at https://data.mendeley.com/datasets/vw2bb5jy99/2 via 'trained_models' and output plots can be found in the 'output' directory.

License

This project is covered under the MIT License.

Contact

Please direct any questions to [e-mail]