Skip to main content

Official repo for Foundation Models for Quantitative Biomarker Discovery in Cancer Imaging [INSERT DOI]

Project description



Build Status Python Version Dependencies Status Code style: black Security: bandit Pre-commit Semantic Versions License Coverage

This is the the official repository for the paper:
"Foundation Models for Quantitative Biomarker Discovery in Cancer Imaging "
Suraj Pai, Dennis Bontempi, Vasco Prudente, Ibrahim Hadzic, Mateo Sokač, Tafadzwa L. Chaunzwa, Simon Bernatz, Ahmed Hosny, Raymond H Mak, Nicolai J Birkbak, Hugo JWL Aerts


NOTE For detailed documentation check our website


Overview

A foundation model, based on a deep convolutional encoder architecture, was pre-trained by contrasting volumes with and without lesions (see a. in Figure). The foundation model was then used to extract biomarkers and evaluated for three classification tasks on diverse datasets. (see b.). We adapted foundation model to specific use cases by extracting features or through fine-tuning (see c.). Performance of the foundation models was compared against conventional supervised implementations, trained from scratch (left) and fine-tuned from a different task (right) (in d). The comparison was made through several criteria for different use cases, including quantitative performance, stability, and biological analysis. Biological, clinical, and stability analyses are limited to use case 2 due to the availability of associated data.

Quick Start (in under 15 mins!)

We provide two notebooks that can be run in google colab to promote the use of our model by the community,

  1. A reproducibility notebook that allows users to recreate our analysis for NSCLC prognostication, from downloading the data to obtaining results matching our manuscript.

  2. A bring your own use case notebook that shows you how you can use the foundation model for your data.

You are not limited to using Colab and you can run our model on your data locally in only 3 steps!

Step 1: Install all our dependencies:

pip install foundation-cancer-image-biomarker --pre

Step 2: Generate a CSV file with the path to your images and seed points (in physical coordinates),

image_path coordX coordY coordZ
/content/data/dummy_0.nii.gz 55.0 119.0 27.0

Step 3: Run this in your code environment,

from fmcib.run import get_features

feature_df = get_features("csv_path_here")

This will preprocess your data, download our pre-trained model and execute inference on whatever device you have (CPU/GPU) and return a dataframe with the features.

License

This project is licensed under the terms of the MIT license. See LICENSE for more details.

Acknowledgements

Code development, testing, and documentation: Suraj Pai, Ibrahim Hadzic
Framework used for building the code: project-lighter

project-lighter was developed internally within our lab by Ibrahim Hadzic and Suraj Pai.

Disclaimer

The code and data of this repository are provided to promote reproducible research. They are not intended for clinical care or commercial use. The software is provided "as is", without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose and noninfringement. In no event shall the authors or copyright holders be liable for any claim, damages or other liability, whether in an action of contract, tort or otherwise, arising from, out of or in connection with the software or the use or other dealings in the software.

Code Citation

Will be updated to reflect a Zenodo DOI upon publication

@misc{foundation-cancer-image-biomarker,
  author = {AIM-Harvard},
  title = {Official repo for "Foundation Models for Quantitative Biomarker Discovery in Cancer Imaging"},
  year = {2023},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/AIM-Harvard/foundation-cancer-image-biomarker}}
}

Project details


Release history Release notifications | RSS feed

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

Built Distribution

File details

Details for the file foundation_cancer_image_biomarker-0.0.1a20.tar.gz.

File metadata

File hashes

Hashes for foundation_cancer_image_biomarker-0.0.1a20.tar.gz
Algorithm Hash digest
SHA256 9a75b6b3785c94e087b696ad83529929dd6901b73c7233c1d9e7972f4dfdb09b
MD5 8fb0602270de134c3508277c0262be7e
BLAKE2b-256 4ae3eb2194a80328f29a3fac57d8d6bb9a1b6c32bf3a5dfe9bb400dab0e4bead

See more details on using hashes here.

File details

Details for the file foundation_cancer_image_biomarker-0.0.1a20-py3-none-any.whl.

File metadata

File hashes

Hashes for foundation_cancer_image_biomarker-0.0.1a20-py3-none-any.whl
Algorithm Hash digest
SHA256 9e6a808dddbeda0807af7e8016ba57eb3c164b7f91ed264a706c93d2c7c89687
MD5 071dc5f2d675e2c8edad6b18b45a109d
BLAKE2b-256 e03a3e895b27aae43f51e81872d9f7438b5d688e5bd9ede63f8582f677fe7eae

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page