Automatic Renal Segmentation for MR Urography Using 3D-GrabCut and Random Forests
This is the implementation of the algorithm described in the following paper:
The source code is located in GitHub repository:
https://github.com/umityoruk/renal-segmentation
The easiest way to run/test this algorithm is the docker image located in Docker Hub:
https://hub.docker.com/r/umityoruk/renal-segmentation/
Run the docker image using:
docker run -it --rm -v /path/to/local/dir:/data -p 8888:8888 umityoruk/renal-segmentation
The image starts the jupyter notebook server at port 8888. You can access the notebook by using the link provided in the terminal. The path /path/to/local/dir is a directory on the host machine that is mounted as /data on the docker container. If you put your dicom images in this directory, you can access them from the Jupyter Notebook running inside the docker container.
See "/Notebook/Automatic_Segmentation_Example.ipynb" for usage examples.
To stop the image simply hit Ctrl-C twice in the terminal.
If you want to process dicom images directly without using the notebook you can run the python command directly:
docker run --rm -v /path/to/local/dir:/data umityoruk/renal-segmentation "python renalSegment.py /data/DicomIn /data/DicomOut"
The example above assumes that the dicom images are stored in /path/to/local/dir/DicomIn and the output folder DicomOut is the last parameter to the renalSegment script.