DemonsRegistration

Diffeomorphic demons registration between two volumes.

DemonsRegistration -f <fixed> -m <moving> [options]

DESCRIPTION

DemonsRegistration performs diffeomorphic demons-based non-linear registration between a fixed and a moving image. The algorithm iteratively estimates a dense deformation field that warps the moving image to match the fixed image, using the demons force with Gaussian smoothing of the deformation and update fields.

Multiple update rules are supported: diffeomorphic (default), additive, and compositive. Gradient computation can be symmetrized, computed from the fixed image, the warped moving image, or the mapped moving image. Multi-resolution registration is controlled by specifying iteration counts per level.

Optional masks can restrict the registration to regions of interest, and an initial deformation field or affine transform can be provided as a starting point.

OPTIONS

-f, --fixed-image image

Fixed (target) image. Required.

-m, --moving-image image

Moving (source) image. Required.

--fixed-mask image

Mask for the fixed image (restrict registration to masked region).

--moving-mask image

Mask for the moving image.

-b, --input-field field

Initial deformation field to start from.

-p, --input-transform xfm

Initial affine transform to apply before deformable registration.

-o, --output-image image

Output the warped moving image.

-O, --outputDef-field field

Output the final deformation field.

-J, --output-jacobian image

Output the Jacobian determinant map of the deformation.

-r, --true-field field

True deformation field for validation (computes error metrics).

-i, --num-iterations UINTxUINTxUINT

Number of iterations per resolution level, specified as a product string (default: 15x10x5).

-s, --def-field-sigma float

Gaussian smoothing sigma for the deformation field (default: 1.5).

-g, --up-field-sigma float

Gaussian smoothing sigma for the update field (default: 0.0).

-l, --max-step-length float

Maximum step length for the demons force (default: 2.0).

-a, --update-rule int

Update rule: 0 = diffeomorphic (default), 1 = additive, 2 = compositive.

-t, --gradient-type int

Gradient computation type: 0 = symmetrized (default), 1 = fixed image, 2 = warped moving image, 3 = mapped moving image.

-e, --use-histogram-matching

Use histogram matching to normalize intensities before registration.

-d, --verbose level

Set debug/verbosity level.

EXAMPLES

# Basic registration
DemonsRegistration -f fixed.mnc -m moving.mnc \
    -o warped.mnc -O deformation.mnc

# Multi-resolution with histogram matching
DemonsRegistration -f fixed.mnc -m moving.mnc \
    -o warped.mnc -O deformation.mnc \
    -i 30x20x10 -e

# Using an initial affine transform with masks
DemonsRegistration -f fixed.mnc -m moving.mnc \
    --fixed-mask mask.mnc -p initial.xfm \
    -o warped.mnc -O deformation.mnc -s 2.0

AUTHOR

Tom Vercauteren - INRIA & Mauna Kea Technologies, with additions by Vladimir S. Fonov.

COPYRIGHTS

Copyright © 2008 by Tom Vercauteren, INRIA & Mauna Kea Technologies

SEE ALSO

LogDomainDemonsRegistration , minctracc , itk_resample