pipeline_deface.pl

defacing pipeline for anonymizing MRI data

pipeline_deface.pl <T1w> [T2w] [PDw] <oT1w> [oT2w] [oPDw] [options]

DESCRIPTION

pipeline_deface.pl is a wrapper script for the BIC defacing pipeline that removes identifying facial features from MRI volumes to protect subject privacy. It supports T1-weighted, T2-weighted, and proton-density (PD) volumes, and can process them simultaneously to ensure consistent defacing across modalities.

The pipeline performs linear and optionally nonlinear registration to a reference model, identifies the facial region, and replaces it with either zeroed voxels or a random distortion grid. Intensity normalization can be disabled with –no-int-norm, and a visible watermark can be applied to the defaced output. The BEaST brain masking library can be specified for improved brain extraction.

OPTIONS

--verbose

Print progress information during processing.

--clobber

Overwrite output files if they already exist.

--dual-echo

Indicate that the T2 and PD inputs are from a dual-echo acquisition.

--model-dir <dir>

Path to the directory containing the reference model files.

--model <name>

Base name of the reference model to use for registration.

--nonlinear

Use nonlinear registration for more accurate facial region identification.

--no-int-norm

Disable intensity normalization of the input volumes.

--watermark

Apply a visible watermark to the defaced output volumes.

--t1w-xfm <xfm>

Pre-computed linear transform for the T1-weighted volume to stereotaxic space.

--t2w-xfm <xfm>

Pre-computed linear transform for the T2-weighted volume to stereotaxic space.

--pdw-xfm <xfm>

Pre-computed linear transform for the PD-weighted volume to stereotaxic space.

--brain-mask <file>

Pre-computed brain mask to use instead of generating one.

--keep-tmp

Do not delete temporary files after processing.

--keep-real-range

Preserve the original real value range in the output MINC files.

--output <dir>

Directory where output files will be written.

--beastlib <dir>

Path to the BEaST brain masking library directory.

--3t

Indicate that the data was acquired on a 3 Tesla scanner, adjusting processing parameters accordingly.

EXAMPLES

Deface a T1-weighted volume:

pipeline_deface.pl input_t1.mnc defaced_t1.mnc

Deface T1 and T2 volumes with nonlinear registration:

pipeline_deface.pl input_t1.mnc input_t2.mnc \
  defaced_t1.mnc defaced_t2.mnc --nonlinear

Deface with watermark and custom model:

pipeline_deface.pl input_t1.mnc defaced_t1.mnc \
  --model-dir /opt/minc/share/model --model icbm152 \
  --watermark --clobber

AUTHOR

Vladimir S. Fonov - McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University.

COPYRIGHTS

Copyright © McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University.

SEE ALSO

deface_minipipe.pl