ISMRM 2015 Tractography Challenge Scoring system
This system contains the scripts and tools that can be used to recreate the results of the ISMRM 2015 Tractography Challenge and to evaluate new datasets.
The release used to produce results for the website and paper is archived on Zenodo .
Configuration and installation
Make sure the "pip" version is recent enough. You can update it with
pip install -U pip
pip install -r requirements.txt pip install -r requirements_additional.txt
You can then clone or download the scoring system. Once cloned or
downloaded, go inside the
directory, and run
python setup.py build_all
Once those steps are all done, the system is configured.
Fetching the Ground Truth Dataset
To be able to run the scoring system, a directory containing the ground truth dataset is needed. It can be downloaded from the Tractometer website.
Scoring a tractogram
Before trying to run the script, the terminal needs to be configured
to correctly find the code. Suppose the code is cloned in a directory
CODE_DIR/ismrm_2015_tractography_challenge_scoring, run the following
Once the ground truth dataset is unarchived (for example, to the
scoring_data directory, one needs to create a directory where
all results will be saved. Let's call it
results for now.
Then, an example call to the scoring system will be
./scripts/score_tractogram.py YOUR_TRACTOGRAM_FILE scoring_data/ results/
YOUR_TRACTOGRAM_FILE is replaced with the path of the
tractogram file that will be scored.
Additional flags use to control the saving behavior of the script are
score_tractogram.py -h to get the list of such