#include <GenericRegistrationFilter.h>

Inheritance diagram for mirtk::GenericRegistrationFilter:

Collaboration diagram for mirtk::GenericRegistrationFilter:

Classes
struct	ConstraintInfo

struct	DisplacementInfo
	Structure storing information about cached displacements. More...

struct	ImageSimilarityInfo

struct	PointSetConstraintInfo

struct	PointSetDistanceInfo

struct	PointSetOutputInfo
	Structure storing information about transformed output point set. More...

struct	TransformationInfo
	Structure storing information about transformation instance. More...

Public Types
typedef RegisteredImage::DisplacementImageType	DisplacementImageType
	Type of cached displacement field.

typedef Array< ResampledImageType >	ResampledImageList
	List type storing images for one resolution pyramid level.

typedef RegisteredImage::InputImageType	ResampledImageType
	Type of resolution pyramid images.

typedef ResampledImageType::VoxelType	VoxelType
	Scalar type of resolution pyramid images.

Public Member Functions
void	AddInput (const BaseImage *)

void	AddInput (vtkPointSet *, double=.0)

double	BackgroundValue (int) const
	Get background value of n-th input image (after registration done)

virtual void	Clear ()
	Reset filter settings and input.

void	ExtrapolationMode (enum ExtrapolationMode)
	Set common/default extrapolation mode for all input images.

enum ExtrapolationMode	ExtrapolationMode (int=-1) const
	Get extrapolation mode of n-th input image or default mode, respectively.

void	ExtrapolationMode (int, enum ExtrapolationMode)
	Set extrapolation mode of n-th input image.

	GenericRegistrationFilter ()
	Constructor.

virtual void	GuessParameter ()

void	Input (const BaseImage , const BaseImage )
	Set input images of the registration filter.

void	Input (int, const BaseImage **)
	Set input images of the registration filter.

template<class TVoxel >
void	Input (int, const GenericImage< TVoxel > **)
	Set input images of the registration filter.

void	Input (vtkPointSet , vtkPointSet , double=.0, double=1.0)

void	Input (int, vtkPointSet *, double =NULL)

void	InterpolationMode (enum InterpolationMode)
	Set common/default interpolation mode for all input images.

void	InterpolationMode (int, enum InterpolationMode)
	Set interpolation mode of n-th input image.

enum InterpolationMode	InterpolationMode (int=-1) const
	Get interpolation mode of n-th input image or default mode, respectively.

bool	IsFixedImage (int) const
	Determine whether the specified input image remains untransformed.

bool	IsFixedPointSet (int) const
	Determine whether the specified input point set remains untransformed.

bool	IsMovingImage (int) const
	Determine whether the specified input image will be transformed.

bool	IsMovingPointSet (int) const
	Determine whether the specified input point set will be transformed.

bool	IsSourceImage (int) const

bool	IsSourcePointSet (int) const

bool	IsTargetImage (int) const

bool	IsTargetPointSet (int) const

virtual Transformation *	MakeInitialGuess ()
	Make an initial guess of the (global) output transformation.

	mirtkPublicAggregateMacro (const Transformation, InitialGuess)
	Initial guess of optimal transformation.

	mirtkPublicAggregateMacro (const Transformation, TargetTransformation)
	Target transformation to be approximated by output transformation.

	mirtkPublicAggregateMacro (BinaryImage, Domain)
	Mask which defines where to evaluate the energy function.

	mirtkPublicAttributeMacro (int, NumberOfLevels)
	Number of resolution levels.

	mirtkPublicAttributeMacro (int, FinalLevel)
	Level at which to stop multi-resolution optimization.

	mirtkPublicAttributeMacro (MFFDMode, MultiLevelMode)
	Multi-level transformation mode.

	mirtkPublicAttributeMacro (Array< enum TransformationModel >, TransformationModel)
	Transformation model.

	mirtkPublicAttributeMacro (enum InterpolationMode, DefaultInterpolationMode)
	Default image interpolation mode.

	mirtkPublicAttributeMacro (enum ExtrapolationMode, DefaultExtrapolationMode)
	Default image extrapolation mode.

	mirtkPublicAttributeMacro (double, MaxRescaledIntensity)

	mirtkPublicAttributeMacro (bool, PrecomputeDerivatives)
	Whether to precompute image derivatives or compute them on the fly.

	mirtkPublicAttributeMacro (enum SimilarityMeasure, SimilarityMeasure)
	Default similarity measure.

	mirtkPublicAttributeMacro (enum PointSetDistanceMeasure, PointSetDistanceMeasure)
	Default polydata distance measure.

	mirtkPublicAttributeMacro (enum OptimizationMethod, OptimizationMethod)
	Optimization method.

	mirtkPublicAttributeMacro (bool, NormalizeWeights)
	Normalize weights of energy function terms.

	mirtkPublicAttributeMacro (string, TargetTransformationErrorName)
	Name of target transformation MSDE term.

	mirtkPublicAttributeMacro (double, TargetTransformationErrorWeight)
	Weight of target transformation MSDE term.

	mirtkPublicAttributeMacro (bool, MergeGlobalAndLocalTransformation)
	Whether to merge initial global transformation into (local) transformation.

	mirtkPublicAttributeMacro (bool, RegisterX)
	Whether to allow x coordinate transformation.

	mirtkPublicAttributeMacro (bool, RegisterY)
	Whether to allow y coordinate transformation.

	mirtkPublicAttributeMacro (bool, RegisterZ)
	Whether to allow z coordinate transformation.

	mirtkPublicAttributeMacro (bool, DirichletBoundaryCondition)

	mirtkPublicAttributeMacro (bool, AdaptiveRemeshing)
	Whether to adaptively remesh surfaces before each gradient step.

int	NumberOfImages () const
	Number of input images.

int	NumberOfPointSets () const
	Number of input point sets.

int	NumberOfRequiredImages () const

int	NumberOfRequiredPointSets () const

virtual ParameterList	Parameter () const
	Get parameters as key/value as string map.

virtual ParameterList	Parameter (int) const
	Get parameters as key/value as string map.

virtual void	ParseEnergyFormula (int=-1, int=-1, int=-1)

virtual bool	Read (istream &, bool=false)
	Read registration parameters from input stream.

virtual void	Reset ()
	Reset filter settings, but keep input.

virtual void	Run ()
	Run the multi-level registration.

virtual bool	Set (const char , const char )
	Set named parameter from value as string.

virtual bool	Set (const char , const char , int)
	Set named parameter from value as string.

virtual void	TransformationModel (enum TransformationModel)
	Set (single) transformation model.

virtual void	Write (const char *) const
	Write registration parameters to file.

virtual	~GenericRegistrationFilter ()
	Destructor.

Public Member Functions inherited from mirtk::RegistrationFilter
void	Output (Transformation **)
	Set pointer to output transformation.

Transformation *	Output ()
	Get output transformation.

virtual bool	Read (const char *, bool=false)
	Read registration parameters from input stream.

virtual	~RegistrationFilter ()=0
	Destructor.

Public Member Functions inherited from mirtk::Observable
void	AddObserver (Observer &)
	Add observer.

void	Broadcast (Event, const void *=NULL)
	Broadcast event to observers.

void	ClearObservers ()
	Delete all observers.

void	DeleteObserver (Observer &)
	Delete observer.

void	NotifyObservers (Event, const void *=NULL)
	Notify all observers about given event if this object has changed.

int	NumberOfObservers () const
	Number of current observers.

virtual	~Observable ()
	Destructor.

Public Member Functions inherited from mirtk::Object
virtual const char *	NameOfClass () const =0
	Get name of class, which this object is an instance of.

bool	Parameter (const ParameterList &)
	Set parameters from name/value pairs.

virtual	~Object ()
	Destructor.

Protected Member Functions
virtual void	AddImageSimilarityTerm ()

virtual void	AddPenaltyTerm ()

virtual void	AddPointSetConstraintTerm ()

virtual void	AddPointSetDistanceTerm ()

virtual void	ApplyInitialGuess ()
	Initialize transformation parameters using provided initial guess.

bool	AtFinalLevel () const
	Whether current level is the final resolution level.

bool	AtInitialLevel () const
	Whether current level is the initial resolution level.

virtual Vector3D< double >	AverageOutputResolution (int=-1) const
	Average resolution of (untransformed) target data sets at given level.

virtual void	Finalize ()
	Finalize registration at current resolution.

virtual struct ImageAttributes	ImageAttributes (int, int=-1) const
	Get attributes of n-th input image at specified resolution level.

virtual void	Initialize ()
	Initialize registration at current resolution.

virtual void	InitializeEnergy ()
	Initialize registration energy of registration at current resolution.

virtual void	InitializeOptimizer ()
	Initialize optimizer used to solve registration problem.

virtual void	InitializeOutput ()
	Initialize transformation for sub-registration at current resolution.

virtual void	InitializePointSets ()
	Remesh/-sample input point sets.

virtual void	InitializePyramid ()
	Initialize image resolution pyramid.

virtual void	InitializeStatus (HomogeneousTransformation *)
	Initialize status of linear parameters.

virtual void	InitializeStatus (FreeFormTransformation *)
	Initialize status of FFD parameters.

virtual void	InitializeStatus ()
	Initialize status of transformation parameters.

virtual void	InitializeTransformation ()
	Initialize new transformation instance.

virtual void	MultiResolutionOptimization ()
	Run multi-resolution registration.

virtual int	NumberOfFrames (double =NULL, double =NULL, double *=NULL) const
	Determine number of temporal frames and temporal sampling attributes.

RegisteredPointSet *	OutputPointSet (int, double, TransformationInfo)

virtual Transformation *	OutputTransformation (TransformationInfo)
	Get (partial/inverse) output transformation.

void	PreUpdateCallback (bool)
	Callback function called by _Energy->Update(bool)

virtual struct ImageAttributes	RegistrationDomain (int=-1) const
	Common attributes of (untransformed) input target data sets at specified resolution level.

virtual void	SetInputOf (RegisteredImage *, const struct ImageAttributes &, int, TransformationInfo)
	Initialize output image corresponding to registered (transformed) image.

virtual enum TransformationType	TransformationType ()
	Type of output transformation of sub-registration at current resolution.

Protected Member Functions inherited from mirtk::RegistrationFilter
bool	Output (Transformation *)
	Set current output transformation.

	RegistrationFilter ()
	Constructor.

Protected Member Functions inherited from mirtk::Observable
	Observable ()
	Default constructor.

	Observable (const Observable &)
	Copy constructor.

Observable &	operator= (const Observable &)
	Assignment operator.

Protected Member Functions inherited from mirtk::Object
template<typename... Args>
void	Throw (ErrorType err, const char *func, Args... args) const

Protected Attributes
Array< double >	_Background
	Image background value.

Array< double >	_Blurring [MAX_NO_RESOLUTIONS]
	Image blurring value.

int	_Centering [MAX_NO_RESOLUTIONS]
	Whether to center foreground (if applicable)

Array< Point >	_Centroid
	Centroids of images.

Array< ConstraintInfo >	_ConstraintInfo
	Parsed constraint(s)

int	_CropPadFFD
	Whether to crop/pad FFD lattice.

bool	_CropPadImages
	Whether to crop/pad input images.

int	_CurrentLevel
	Current resolution level.

enum TransformationModel	_CurrentModel
	Current transformation model.

double	_DefaultBackground
	Default background value.

Array< DisplacementInfo >	_DisplacementInfo
	Info of cached displacement fields.

bool	_DownsampleWithPadding

RegistrationEnergy	_Energy
	Registration energy.

string	_EnergyFormula
	Registration energy formula as string.

EventDelegate	_EventDelegate
	Forwards optimization events to observers.

Array< enum ExtrapolationMode >	_ExtrapolationMode
	Extrapolation mode of each input image.

Array< ResampledImageList >	_Image
	Resolution pyramid.

Array< ImageSimilarityInfo >	_ImageSimilarityInfo
	Parsed similarity measure(s)

Array< const BaseImage * >	_Input
	Input images.

Array< enum InterpolationMode >	_InterpolationMode
	Interpolation mode of each input image.

Array< BinaryImage * >	_Mask
	Domain on which to evaluate similarity.

double	_MaxControlPointSpacing [MAX_NO_RESOLUTIONS][4]
	Control point spacing for FFDs.

Array< double >	_MaxEdgeLength [MAX_NO_RESOLUTIONS]
	Maximum edge length in mm.

double	_MinControlPointSpacing [MAX_NO_RESOLUTIONS][4]
	Control point spacing for FFDs.

Array< double >	_MinEdgeLength [MAX_NO_RESOLUTIONS]
	Minimum edge length in mm.

LocalOptimizer *	_Optimizer
	Used optimizer.

ParameterList	_Parameter [MAX_NO_RESOLUTIONS]

Array< Array< void * > >	_PointSet
	Unused.

Array< PointSetConstraintInfo >	_PointSetConstraintInfo
	Unused.

Array< PointSetDistanceInfo >	_PointSetDistanceInfo
	Unused.

Array< void * >	_PointSetInput
	Unused.

Array< void * >	_PointSetOutput
	Unused.

Array< PointSetOutputInfo >	_PointSetOutputInfo
	Unused.

Array< double >	_PointSetTime
	Unused.

RegistrationEnergy::PreUpdateFunctionType	_PreUpdateDelegate
	Delegate of pre-update function.

struct ImageAttributes	_RegistrationDomain

Array< Vector3D< double > >	_Resolution [MAX_NO_RESOLUTIONS]
	Image resolution in mm.

Point	_SourceOffset
	Source origin offset.

bool	_Subdivide [MAX_NO_RESOLUTIONS][4]
	Whether to subdivide FFD.

Point	_TargetOffset
	Target origin offset.

Transformation *	_Transformation
	Current estimate.

Array< TransformationInfo >	_TransformationInfo
	Meta-data of partial transformation.

Array< Transformation * >	_TransformationInstance
	Partial transformations.

int	_UseGaussianResolutionPyramid
	Whether resolution levels correspond to a Gaussian pyramid.

Additional Inherited Members
Static Public Member Functions inherited from mirtk::Object
static const char *	NameOfType ()
	Get name of this class type.

Static Protected Member Functions inherited from mirtk::Object
template<typename... Args>
static void	ThrowStatic (ErrorType err, const char cls, const char func, Args... args)

Detailed Description

Generic registration filter

Definition at line 72 of file GenericRegistrationFilter.h.

Member Function Documentation

§ AddImageSimilarityTerm()

virtual void mirtk::GenericRegistrationFilter::AddImageSimilarityTerm ( )

protectedvirtual

Instantiate new image similarity term for energy function

Note: The individual energy terms are destroyed by the energy function!

§ AddInput() [1/2]

void mirtk::GenericRegistrationFilter::AddInput ( const BaseImage * )

Add filter input image

For registration filters which support multiple target/source images such as multiple channels and/or frames of a temporal image sequence. How the transformation is applied to which inputs depends on the respective transformation model and image registration method.

§ AddInput() [2/2]

void mirtk::GenericRegistrationFilter::AddInput	(	vtkPointSet *	,
		double	= `.0`
	)

Add filter input point set

Note: Use only when MIRTK_Registration_WITH_PointSet is 1.

§ AddPenaltyTerm()

virtual void mirtk::GenericRegistrationFilter::AddPenaltyTerm ( )

protectedvirtual

Instantiate new regularization term for energy function

Note: The individual energy terms are destroyed by the energy function!

§ AddPointSetConstraintTerm()

virtual void mirtk::GenericRegistrationFilter::AddPointSetConstraintTerm ( )

protectedvirtual

Instantiate new point set constraint term for energy function

Note: The individual energy terms are destroyed by the energy function!

§ AddPointSetDistanceTerm()

virtual void mirtk::GenericRegistrationFilter::AddPointSetDistanceTerm ( )

protectedvirtual

Instantiate new point set distance term for energy function

Note: The individual energy terms are destroyed by the energy function!

§ GuessParameter()

virtual void mirtk::GenericRegistrationFilter::GuessParameter ( )

virtual

Guess proper setting for any yet unset parameter

Note: Optional to call explicitly as this method is called by Run.

§ Input() [1/2]

void mirtk::GenericRegistrationFilter::Input	(	vtkPointSet *	,
		vtkPointSet *	,
		double	= `.0`,
		double	= `1.0`
	)

Set input point sets of the registration filter

Note: Use only when MIRTK_Registration_WITH_PointSet is 1.

§ Input() [2/2]

void mirtk::GenericRegistrationFilter::Input	(	int	,
		vtkPointSet **	,
		double *	= `NULL`
	)

Set input point set of the registration filter

Note: Use only when MIRTK_Registration_WITH_PointSet is 1.

§ IsSourceImage()

bool mirtk::GenericRegistrationFilter::IsSourceImage ( int ) const

Determine whether the specified input image will be transformed by the forward transformation or part of it such as in case of a symmetric energy.

§ IsSourcePointSet()

bool mirtk::GenericRegistrationFilter::IsSourcePointSet ( int ) const

Determine whether the specified input point set either remains untransformed, or is transformed by the inverse transformation or a part of it such as in case of an inverse consistent and/or symmetric energy function.

§ IsTargetImage()

bool mirtk::GenericRegistrationFilter::IsTargetImage ( int ) const

Determine whether the specified input image either remains untransformed, or is transformed by the inverse transformation or a part of it such as in case of an inverse consistent and/or symmetric energy function.

§ IsTargetPointSet()

bool mirtk::GenericRegistrationFilter::IsTargetPointSet ( int ) const

Determine whether the specified input point set will be transformed by the forward transformation or part of it such as in case of a symmetric energy.

§ mirtkPublicAttributeMacro() [1/2]

mirtk::GenericRegistrationFilter::mirtkPublicAttributeMacro	(	double	,
		MaxRescaledIntensity
	)

Rescale input image intensities from [imin, imax] to [0, omax]

Here [imin, imax] is the intensity range of the respective input image excluding the background (if specified) and omax is the maximum intensity of the rescaled intensities. This maximum value influences the relative weighting of image similarity gradient vs. the gradient of constraint terms such as the bending energy.

This normalization ensures that the magnitude of image derivatives used for energy gradient computation have comparable magnitude in order to prevent one image to have stronger influence than another. This is more important for symmetric, inverse consistent, or multi-modal settings.

By default, this parameter is set to inf and no rescaling is done.

§ mirtkPublicAttributeMacro() [2/2]

mirtk::GenericRegistrationFilter::mirtkPublicAttributeMacro	(	bool	,
		DirichletBoundaryCondition
	)

Enforce Dirichlet boundary condition on FFD transformations

When this option is enabled, the status of control points at the boundary of the finite FFD lattice is set to Passive and the parameters of these control points set to zero. This is always the case for FFDs whose parameters are control point displacements. In case of FFDs parameterized by (stationary) velocities, the default extrapolation mode is nearest neighbor, however, and a layer of passive CPs with constant value is needed if the velocity should be forced to zero outside the finite domain on which the velocity field is defined. Alternatively, FFD extrapolation mode "Const" can be used.

§ NumberOfRequiredImages()

int mirtk::GenericRegistrationFilter::NumberOfRequiredImages ( ) const

Number of required input images

Note: Only valid after ParseEnergyFormula has been called!

§ NumberOfRequiredPointSets()

int mirtk::GenericRegistrationFilter::NumberOfRequiredPointSets ( ) const

Number of required input point sets

Note: Only valid after ParseEnergyFormula has been called!

§ OutputPointSet()

RegisteredPointSet* mirtk::GenericRegistrationFilter::OutputPointSet	(	int	,
		double	,
		TransformationInfo
	)

protected

Get (new) registered output point set

Note: Use only when MIRTK_Registration_WITH_PointSet is 1.

§ ParseEnergyFormula()

virtual void mirtk::GenericRegistrationFilter::ParseEnergyFormula	(	int	= `-1`,
		int	= `-1`,
		int	= `-1`
	)

virtual

Parse registration energy function

Note: Optional to call explicitly as this method is called by GuessParameter.

Member Data Documentation

§ _DownsampleWithPadding

bool mirtk::GenericRegistrationFilter::_DownsampleWithPadding

protected

Whether to take background into account during initialization of the image pyramid

Definition at line 359 of file GenericRegistrationFilter.h.

§ _Parameter

ParameterList mirtk::GenericRegistrationFilter::_Parameter[MAX_NO_RESOLUTIONS]

protected

Parameters not considered by the registration filter itself

These parameters are passed on to the sub-modules of the registration filter such as the image similarity measure(s), the registration constraint(s), and the optimizer. This way, the registration filter does not need to know which parameters its sub-modules accept and is better decoupled from the implementation of the respective image similarities, constraints, and optimizers.

Definition at line 395 of file GenericRegistrationFilter.h.

§ _RegistrationDomain

struct ImageAttributes mirtk::GenericRegistrationFilter::_RegistrationDomain

protected

Common attributes of (untransformed) input target data sets

These attributes are in particular used to initialize the control point grid of a free-form deformation such that the grid is large enough to be valid for every target point for which the transformation will be evaluated. It is therefore computed from the attributes of the input data sets rather than the downsampled data. If a _Domain mask is set, the attributes of this mask are copied instead.

Definition at line 331 of file GenericRegistrationFilter.h.

The documentation for this class was generated from the following file:

GenericRegistrationFilter.h

Classes

Public Types

Public Member Functions

Protected Member Functions

Protected Attributes

Additional Inherited Members

Detailed Description

Member Function Documentation

§ AddImageSimilarityTerm()

§ AddInput() [1/2]

§ AddInput() [2/2]

§ AddPenaltyTerm()

§ AddPointSetConstraintTerm()

§ AddPointSetDistanceTerm()

§ GuessParameter()

§ Input() [1/2]

§ Input() [2/2]

§ IsSourceImage()

§ IsSourcePointSet()

§ IsTargetImage()

§ IsTargetPointSet()

§ mirtkPublicAttributeMacro() [1/2]

§ mirtkPublicAttributeMacro() [2/2]

§ NumberOfRequiredImages()

§ NumberOfRequiredPointSets()

§ OutputPointSet()

§ ParseEnergyFormula()

Member Data Documentation

§ _DownsampleWithPadding

§ _Parameter

§ _RegistrationDomain