ruefene / SkullStrippingPipeline

An PyRaDiSe-based auto-segmentation example using Docker and providing a web interface for interaction.

SkullStrippingPipeline

This repository contains a Docker-based auto-segmentation pipeline example relying on the PyRaDiSe Python package.

Installation

Clone the repository by typing the following command in your terminal:

git clone https://github.com/ruefene/SkullStrippingPipeline.git

Navigate to the data/model directory and download the example model file from the PyRaDiSe example data repository:

cd SkullStrippingPipeline/data/model
curl -LJO https://github.com/ruefene/pyradise-example-data/raw/main/model/model.pth

Then, go to the repository folder and build the Docker image by typing the following command:

cd SkullStrippingPipeline
docker build -t ch.unibe.segmentation.skullstripping .

Usage

To run the pipeline, type the following command in your terminal:

docker run -p 4000:5000 --name SkullStripper --gpus all ch.unibe.segmentation.skullstripping 

After starting the container, you can access the web interface by typing the following URL in your browser:

http://localhost:4000

Run without Docker

For running this example without using Docker you need to set several environment variables. The following table shows the required environment variables and their default values:

Variable	Example value	Description
MODEL_DIR_PATH	C:\SkullStrippingPipeline\data\model	The directory where the model is stored.
INPUT_DATA_DIR	C:\SkullStrippingPipeline\data\input	The directory where the input data is uploaded to.
OUTPUT_DATA_DIR	C:\SkullStrippingPipeline\data\output	The directory where the output data will be stored before download.
SCRATCH_DATA_DIR	C:\SkullStrippingPipeline\data\scratch	The directory where the temporary data gets stored.

License

This project is licensed under the MIT License - see the LICENSE file for details