Diameter learning

Train on network on the regression of the lumen diameter to obtain the full lumen segmentation.

Requirements

• Python 3.7
• Conda (optional)

Installation

• For conda users:

$ conda env create --name [env-name] --file=conda.yaml

• For others:

$ pip install -r requirements.txt

Structure

• CONTRIBUTING.md: File that set up of the continous integration
• MLproject: File that set up MLexperiments
• LICENSE: File that contains the legal license
• diameter_learning/: Contains the code of your project the structure is similar to MONAI project structure to make it easier to contribute
• scripts/: Directory that contains the entry points of the program
• test/: Directory that contains the tests of this repository

Usage

Obtain data

You can download the data as a zip archive by joining the Carotid Artery Vessel Wall Segmentation Challenge. Once downloaded, you can place get the data in the right folder by using:

• if you have the zip archive with the data:

$ make data_zip ZIP_PATH="[your absolute path to the zip archive]"

• if you have already inflated the zip archive with the data:

$ make data_repo REPO_PATH="[your absolute path to inflated folder]"

Preprocess data

• Once you obtained the data you can preprocess them with the command

$ make preprocess

Run tests

• Run the tests in your environment

$ make test

Launch experiments

• Launch an mlflow experiment with conda

$ mlflow run ./ -e [entry-point]

• Launch an mlflow experiment without conda

$ mlflow run ./ -e [entry-point] --no-conda

To contribute

Please follow the recommendation of the CONTRIBUTING.md

Authors

The following authors contributed :

• Robin Camarasa