Difference between revisions of "Full brain parcellation using Geodesic Information Flow"

Latest revision as of 14:59, 24 November 2015

This pipeline uses Geodesic Information Flow (GIF) in order to propagate an atlas parcellation towards a target subject geometry. It only takes an input T1 image as input and a reference database xml file in a specific format. Simplest usage is the following:

perform_gif_propagation_expanded.py -d <database.xml> -i <T1.nii.gz> -o <output_dir>

The following is a description of the other parameters available from the pipeline:

usage: perform_gif_propagation_expanded.py [-h] -i input_file [input_file ...]
                                           -d database -o output [-n n_procs]
                                           [-u username] [-g]

GIF Propagation

optional arguments:
  -h, --help            show this help message and exit
  -i input_file [input_file ...], --input_file input_file [input_file ...]
                        Input image file to propagate labels in
  -d database, --database database
                        gif-based database xml file describing the inputs
  -o output, --output output
                        output directory to which the gif outputs are stored
  -n n_procs, --n_procs n_procs
                        maximum number of CPUs to be used when using the
                        MultiProc plugin
  -u username, --username username
                        Username to use to submit jobs on the cluster
  -g, --graph           Print a graph describing the node connections

The database is the key argument of the pipeline. It will determine what atlas is to be propagated and the subjects present in the database will influence the propagation.

When in the cmic cluster, the database is located here: /cluster/project0/GIF/template-database-r2.1/db.xml, so you can use the following line:

perform_gif_propagation_expanded.py -d /cluster/project0/GIF/template-database-r2.1/db.xml -i <T1.nii.gz> -o <output_dir>

When in the DRC network, the database is located here: ~toussaint/scratch/data/template-database/db.xml, so you can use the following line:

perform_gif_propagation_expanded.py -d ~toussaint/scratch/data/template-database/db.xml -i <T1.nii.gz> -o <output_dir>

External links:

• [conference paper about the algorithm]<http://doi.org/10.1007/978-3-642-33418-4_33>