LOCALIZATION.md

The Neuroradiology Localization Pipeline

(A Quick Guide)

Installation

1. Clone the create_montage branch of the event_creation repository:

git clone https://github.com/pennmem/event_creation.git -b create_montage

This will clone the repository into a directory named event_creation. 2. Initialize the neurorad submodule:

cd event_creation
git submodule init
git submodule update

3. Create the necessary conda environment:

conda env create -f conda_environment.yml

Usage

To run the localization pipeline:

./submit --localization-only

If the pipeline has been run on a subject, and none of the input files have changed, then this command will not rerun the pipeline. To force the pipeline to rerun, use the following options:

   --force-localization: Force the pipeline to rerun in the absence
                         of a change in the underlying files. Note that
                         this will NOT redo the Dykstra correction

   --force-dykstra:      Force the pipeline to rerun a previously
                         computed Dykstra correction.

Settings

By default, the pipeline script writes everything to /protocols, a directory only writable by the RAM_maint user. If you do not have access to the RAM_maint account and wish to run the pipeline, you will need to change the database directory, which is set in the file submission/configuration/config.yml and is controlled by the db_root setting.