/PCNtoolkit-interface

Code for PCNtoolkit's online presence aka live inference machine

Primary LanguagePythonGNU General Public License v3.0GPL-3.0

Repository for PCNtoolkit Normative Models via Docker Container

View on Docker Hub

Note: The docker-compose YAML file was copied from this GitHub repository and has not been edited at all (yet), meaning it is just a template/placeholder.

This is a python-based container that will install the necessary packages to run pre-trained cortical thickness and subcortical volume normative models and transfer these models to an unseen test set. The models are described in Rutherford et al. The transfer test set is currently a multi-site dataset using public data from OpenNeuro. This build currently does not support using your own transfer data set. This feature (uploading your own dataset) is actively being developed and a future release will allow users an option to input their own transfer data set.

Setup Steps:

  1. Clone this GitHub repository and cd into the cloned repository.

    git clone git@github.com:saigerutherford/PCNtoolkit-interface.git

    cd PCNtoolkit-interface

  2. Pull the Docker image from DockerHub

    docker pull saruther/pcntoolkit-interface

  3. Run the container, and mount the GitHub repository using the Volume (-v) flag. Change the path to match the location of the cloned GitHub repository.

    docker run -v /path/to/PCNtoolkit-interface/repo:/home/jovyan/ pcntoolkit-interface python apply_normative_models.py

After the script has run:

There will be a file called Z_predict.txt that is located in each subdirectory in /models/lifespan_57K_82sites/. This file contains the Z-scores (deviation scores) for all subjects in the test set. There is one subdirectory and corresponding Z_predict.txt file for every model that was run.