registration: Voxel-based affine image registration.

Syntax:
	registration [options] reference.study model.study
or
	registration [options] input.list

where [options] is one or more from:

Flags:
 --copyright
	Print copyright message.
 --copyright-vtk
	CopyrightVTK

	-v, --verbose
		Output diagnostic information (default)
	-q, --quiet
		Suppress all diagnostic output
	-i, --initxlate
		Initially align centers of volumes
	-n, --noswap
		Prevent automatic swapping of model and reference

Input control:
	--crop0 "x0 y0 z0 x1 y1 z1"
		Crop reference image to grid range.
	--crop1 "x0 y0 z0 x1 y1 z1"
		Crop floating image to grid range.
	[--Crop0, --Crop1] "X0 Y0 Z0 X1 Y1 Z1"
		Like --crop0/--crop1, but with real-world coordinates.
	--auto-crop0 level
		Automatically crop reference using threshold level.
	--auto-crop1 level
		Automatically crop floating using threshold level.

Output control:
	-o <dir>, --outlist <dir>
		Output studylist containing registration.
	-t <file>, --time <file>
		Time output file (default: none.)
	-p <file>, --protocol <file>
		Protocol output file (default: none)
Registration control:
	--dofs <intval>
		Number of degrees of freedom. Legal values are
		6 (rigid), 7 (rigid+global scaling), 9 (rigid+scaling),
		or 12 (full affine). Default is 6.
	-s <realval>, --sampling <realval>
		Maximum resampled resolution in mm (default: 1.0)
	-a <realval>, --accuracy <realval>
		Minimum registration stepsize in mm (default: 0.1)
	-e <realval>, --exploration <realval>
		Initial stepsize (exploration) in mm (default: 8.0)
	-f <realval>, --stepfactor <realval>
		Factor between two successive step sizes (default: 0.5)
		Value must be between 0 and 1.
	-m (0|1|2), --metric (0|1|2)
		Metric: 0 = normalized mutual information (default),
		        1 = mutual information,
		        2 = correlation ratio.
		        4 = mean squared difference.
		        5 = cross correlation.

Optimization. One or none of the following (default is best-neighbour-search):
	--powell
		Powell-optimization
	-l, --linesearch
		Linesearch optimizer.

	-L, --Linesearch
		Linesearch optimizer with extensive optimization.


Specialties:
	--threads <intval>|auto
		Run <intval> threads in parallel on multi-processor machines.
		If 'auto' is given as the number of threads the program will
		determine the number of available CPUs installed and will use
		that number as the (maximum) number of threads.