hcag_segmentation_pipeline.pl

run hippocampus and amygdala segmentation pipeline

hcag_segmentation_pipeline.pl <input_t1.mnc> <output_prefix> --model-dir <dir>

DESCRIPTION

hcag_segmentation_pipeline.pl performs automatic segmentation of the hippocampus and amygdala (and optionally other structures depending on the atlas library) using the patch-based segmentation method described in Coupe et al. (2010). The method compares patches from the input image against a library of pre-labelled template images and assigns labels using a non-local means weighting scheme.

The pipeline:

1. Registers the input T1-weighted volume to the model space.
2. Extracts regions of interest around the hippocampus and amygdala.
3. Runs patch-based label fusion using itk_patch_morphology to produce a segmentation.

A model directory containing pre-labelled template libraries must be provided via the --model-dir option.

OPTIONS

--verbose

Print progress information during processing.

--qc

Generate quality control images after segmentation.

--clobber

Overwrite existing output files.

--model-dir <dir>

Path to the directory containing pre-labelled atlas templates. This option is required.

--subject <id>

Subject identifier used for output file naming.

--search <n>

Search radius (in voxels) for patch matching. Default: 2.

--patch <n>

Half-size of the patch used for matching (in voxels). Default: 1.

--exclude <id>

Exclude the specified template from the library (e.g. for leave-one-out validation).

--classes <n>

Number of label classes in the segmentation. Default: 18.

--variant <name>

Name of the label variant to use from the model directory. Default: labels.

--threshold <f>

Confidence threshold for label assignment.

--compare <labels.mnc>

Compare segmentation output against the given ground-truth label volume and report overlap statistics.

EXAMPLES

Segment the hippocampus and amygdala:

hcag_segmentation_pipeline.pl subject_t1.mnc output_prefix \
    --model-dir /opt/models/hcag_library

Segment with custom patch parameters and generate QC:

hcag_segmentation_pipeline.pl --search 3 --patch 2 --qc \
    --clobber subject_t1.mnc seg_output --model-dir /opt/models/hcag_library

Leave-one-out validation:

hcag_segmentation_pipeline.pl --exclude template_01 \
    --compare ground_truth.mnc \
    subject_t1.mnc validation_output --model-dir /opt/models/hcag_library

AUTHOR

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

COPYRIGHTS

Copyright (C) Vladimir S. Fonov and McGill University. Licensed under the terms of the GNU General Public License version 3 or later.

SEE ALSO

itk_patch_morphology(1), patch_segmentation_pipeline.pl(1), mincbeast(1)