CNI - Challenge



How to Submit


MICCAI CNI 2019 Challenge is accepting solutions as Docker images. This allows you to submit a Challenge solution in your choice of open-source language and version. And it enables us to easily run your solution without having to install multiple programs required for execution.

We have provided a wrapper for your code in a GitHub repo, pl-cni_challenge, which can then be linked in Docker to containerise your solution for submission. pl-cni_challenge is an app containing a bare bones example which you can edit to include your code/package. While the wrapper is in Python and currently contains an example also in Python, other languages are possible (a C example will be added soon).

Quick install and setup instructions to create a Docker image are below. See the pl-cni_challenge Github repo README for details on how to execute and populate the bare bones example for your submission.

The cni_challenge.py repo can be found here.
Further information on the Challange data can be found here.
cni_challenge.py is a ChRIS-based application, for more information on the original app, see here

Installation Requirements and Quick Setup

1. Install Python (3.5+) and pip (which is usually installed with Python)
2. Create a GitHub account here if you don't already have one, and install it on your machine.
3. Create a DockerHub account here if you don't already have one.
4. Install the latest Docker (17.04.0+) if you want to test your plugin's Docker image and containers on your local machine.
     To install on Ubuntu 18.04:

     apt-get remove docker docker-engine docker.io
     apt install docker.io

     Otherwise, visit https://docs.docker.com/install/ for installation directions.

5. Fork this pl-cni_challenge repository to your GitHub.
6. Log onto your Docker Hub account and create a new repository with automated build.

     In 'Account Settings' -> 'Linked Accounts', connect your GitHub account to your DockerHub account.

     Then back in your DockerHub home, click the Create Repository+ button. The website page will walk you through setting up the automated build. When prompted for the GitHub repository that you’d like to use for the automated build select the pl-cni_challenge repository that you just forked. Name the Docker repository ${cni_challenge_DockerRepo} and make it Public.

     **It is extremely important that you tag your automatically built docker image with an appropriate version number based on your GitHub tags**.
     Create a new build rule by clicking the BUILD RULES+ button. A good rule good be **Source type:** ``Tag``, **Source:** ``/^[0-9.]+$/`` and **Docker Tag:** ``version-{sourceref}``.

     Click Create && Build button to finish the setup and trigger the automated build.
     For more information on Automated Builds, visit https://docs.docker.com/docker-hub/builds/.

     After ths build has completed, the cni_challenge bare bones example is now available as a Docker image to be pulled from your DockerHub. The link to it will be ${your_Docker_account name}/${cni_challenge_DockerRepo}.

7. There are technical App and Challenge Requirements that need to be followed for us to run and evaluate your submission, see here for details.

How to Submit

To submit, fill out this form, including the above link of your Docker container, and upload it to our CMT Challenge portal here.

An accepted challenge submission must be accompanied by at least one author registered to the CNI-TL Challenge through the MICCAI satellite events registration. However, participants will have the option to submit a short video presentation about their method in lieu of in-person attendance.