This repository is a fork of a machine learning(ML) model made by UCSD. Our goal is to create a reliable ML model that can determine if a patient has COVID-19 based on computed tomography(CT) scans from a sample of patients that have tested positive and patients that have tested negative.
Citations
The additional dataset that we used can be found at https://www.kaggle.com/plameneduardo/sarscov2-ctscan-dataset#. The citations for this dataset are: Angelov, Plamen, and Eduardo Almeida Soares. "EXPLAINABLE-BY-DESIGN APPROACH FOR COVID-19 CLASSIFICATION VIA CT-SCAN." medRxiv (2020). Soares, Eduardo, Angelov, Plamen, Biaso, Sarah, Higa Froes, Michele, and Kanda Abe, Daniel. "SARS-CoV-2 CT-scan dataset: A large dataset of real patients CT scans for SARS-CoV-2 identification." medRxiv (2020). doi: https://doi.org/10.1101/2020.04.24.20078584. Link: https://www.medrxiv.org/content/10.1101/2020.04.24.20078584v2
How to use
Clone the repository
git clone https://github.com/walkerjbuckle/COVID-CT-Starlight-Saviors.git
Create an envirement (using environment.yml or requirements.txt)
-Using Conda environment
Requires Anaconda running python 3.7 or newer
cd COVID-CT-Starlight-Saviors
conda env create -f environment.yml
conda activate starlight
-Using pip requirements.txt
cd COVID-CT-Starlight-Saviors
pip install -r requirements.txt
pip install torch torchvision
The PyTorch neural network models can be downloaded from https://www.kaggle.com/stevengriffin/covidctstarlightsaviors .
Contributors, please install git pre-commit hook
cd COVID-CT-Starlight-Saviors
pip install pre-commit
pre-commit install
pre-commit run --all-files
Who are we?
This project is being done by Interns as part of the MISI Internship program. Interns divided into six teams in order to work on the project. Team 1 was responsible for the command line Linux app and a PyQt cross-platform GUI app. Team 2 was responsible for managing the Git repository, assisting other teams with it, and writing the README. Team 3 was responsible for adding datasets for training, making a script for partitioning the datasets and making an API for data loading. Team 4 was responsible for making an API for training and researching training options. Team 5 was responsible for researching the options for layers and making a new CNN class model. Team 6 was responsible for searching for possible models to apply with transfer learning that have not yet been tried, and to train and validate those models.
Docs folder
The docs folder includes a written paper from each team about how they went about addressing their goals and needs, including any issues or mistakes made along the way. These documents also include the team members as authors for each team and any references needed.