FreeSurfer

Overview

FreeSurfer is a freely available software package developed by investigators at the Athinoula A. Martinos Center for Biomedical Imaging used for a number of procedures including:

1. Creation of computerized models of the brain from magnetic resonance imaging (MRI) data.
2. Processing of functional magnetic resonance imaging (fMRI) data.
3. Measuring a number of morphometric properties of the brain including cortical thickness and regional volumes.
4. Intersubject averaging of structural and functional data using a procedure that aligns individuals based on their cortical folding patterns for optimal alignment of homologous neural regions.

Working with FreeSurfer on a Cluster

FreeSurfer run its process in a non-parallel environment, so there won't be much benefit from a dual/quad/hex core machine for a single case analysis. However a number of FreeSurfer recon-all processes can be started in the same machine, thereby, theoretically reducing the time to analyze a group of cases.

Here's how to run many recon-all processes on Cheaha.

Set up FreeSurfer Environment

Setting up an environment to run FreeSurfer on Cheaha, consists of defining FreeSurfer related variables and their installation paths. Steps 1 and 2 are a one-time set up, ie., the first time you login to Cheaha. Need not set up and source env variables defined in .bashrc from the next time you login to Cheaha (as at login bashrc is auto-sourced).

1. To set up your environment on Cheaha, open $HOME/.bashrc with your favorite text editor such as vim, nano, emacs etc., and copy-paste the contents below:

export FREESURFER_HOME=/share/apps/freesurfer/4.5
source $FREESURFER_HOME/SetUpFreeSurfer.sh
export FSFAST_HOME=$FREESURFER_HOME/fsfast
export FSF_OUTPUT_FORMAT=nii
export MNI_DIR=$FREESURFER_HOME/mni
export FSL_DIR=$HOME/fsl
export PATH=$FREESURFER_HOME/bin:$FSFAST_HOME/bin:$MNI_DIR/bin:$FSL_DIR/bin:$PATH

NOTE: In the above bashrc, FSL_DIR is also defined. FSL is not part of FreeSurfer, but needs to be installed separately. Download FSL from here.

2. Source the above environment variables.

source $HOME/.bashrc

Run example recon_all process

1. Prepare by creating a subjects directory to hold image files in .nii format

mkdir $HOME/subjects

cd $HOME/subjects

2. Get example image by copying the line below

wget http://repo.uabgrid.uab.edu/files/freesurfer/example.nii.gz

gunzip example.nii.gz

Create job submission script

This is a simple SGE script to submit recon-all process. This script first copies your subjects directory from your $HOME containing the MRI data to Cheaha's high performance file system, $UABGRID_SCRATCH, so as to ramp the speed of recon-all process. The recon-all process is then executed on $UABGRID_SCRATCH.

NOTE: For the example run, we've created a subjects directory at $HOME to hold the example.nii image. For your actual run with real images, if you've copied your images to another location, then make sure to change $HOME/subjects (line no 7) accordingly in the script below.

Open a text editor and copy-paste the below contents:

#!/bin/bash

scratch=$UABGRID_SCRATCH
jobs=$scratch/jobs
sub=$scratch/subjects
mkdir $jobs
cp -r $HOME/subjects $scratch/

function run-recon-all() {
cd $scratch
#create submit script for each patient in the subjects directory
for patient in `ls -1 $sub`
do
unset SUBJECTS_DIR
SUBJECTS_DIR=$sub
export SUBJECTS_DIR
cat > $jobs/recon-all-$patient.sh <<EOF 
#!/bin/bash
#$ -S /bin/bash
#$ -cwd
#
#$ -N recon-all-$patient
# Set the hard and soft run time limits (ex: 1hour/58 min)
#$ -l h_rt=30:00:00,s_rt=29:55:00
#$ -j y
#$ -M YOUR_EMAIL_ADDRESS
#$ -m eas
# set up FreeSurfer environment variables

mri_convert -oi -os $SUBJECTS_DIR/$patient $SUBJECTS_DIR/$patient.mgz 

sleep 5

recon-all -sd $SUBJECTS_DIR -s $patient
EOF
done

#submit job script for each patient in the jobs directory to the cluster
pushd $jobs
for script in `ls -1`
do
echo "submitting job $script"
qsub $script
done
popd
}
run-recon-all

Make the above script executable

chmod +x <name-of-script-above>

Then submit that script with

./name-of-script-above

Check on it with qstat.

qstat -u $USER

After the completion of the recon-all process, copy the results and logs back to your $HOME and delete the contents from scratch

cp -r $UABGRID_SCRATCH/subjects $HOME/subjects-results

cp -r $UABGRID_SCRATCH/jobs $HOME/jobs-results

rm -rf $UABGRID_SCRATCH/subjects

rm -rf $UABGRID_SCRATCH/jobs