CorrespondenceModel Class Reference

#include <CorrespondenceModel.h>

Public Types
typedef float	PrecisionType
typedef std::vector< PrecisionType >	SurrogateSignalType

Public Member Functions
	CorrespondenceModel (unsigned int numberOfSurrogateSignalsIn, std::shared_ptr< Transformation > transformationIn)
	~CorrespondenceModel ()
std::shared_ptr< Transformation >	GetTransformationFromSurrogateSignal (const SurrogateSignalType &surrogateSignalIn)
PrecisionType *	GetParameters ()
void	SetParameters (const PrecisionType *parametersIn)
void	SetParameters (const std::vector< nifti_image * > &parameterImagesIn)
PrecisionType	GetMaxParameterLength (PrecisionType *parametersIn)
unsigned int	GetNumberOfParameters ()
void	GetTransformationParameterGradientWRTModelParameters (PrecisionType *transformationGradientIn, const SurrogateSignalType &surrogateSignalIn, PrecisionType *correspondenceModelGradientOut)
void	InitialiseLevel (unsigned int levelIn)
std::shared_ptr< Transformation >	GetTransformation ()
std::vector< nifti_image * >	GetCorrespondenceModelAsImage ()
void	SaveCurrentModelParametersForRecovery ()
void	RecoverSavedModelParameters ()

Private Member Functions
void	ClearRecoveryModelParameters ()

Private Attributes
unsigned int	numberOfSurrogateSignals
	The number of surrogate signals used. More...
unsigned int	numberOfTransformationParameters
	The number of parameters to describe the transformation completely. More...
unsigned int	numberOfModelParameters
	The total number of parameters describing the correspondence model. More...
std::shared_ptr< Transformation >	transform
	The internal transformation object that is used to construct a transformation (copy) from a surrogate signal. More...
PrecisionType *	modelParameters
	Pointer to the model parameters. More...
PrecisionType *	recoveryModelParameters
	Pointer to model parameters that may be recovered. More...

Detailed Description

Class implementing the (respiratory) correspondence model.

Member Typedef Documentation

PrecisionType

typedef float CorrespondenceModel::PrecisionType

SurrogateSignalType

typedef std::vector<PrecisionType> CorrespondenceModel::SurrogateSignalType

Constructor & Destructor Documentation

CorrespondenceModel()

CorrespondenceModel::CorrespondenceModel	(	unsigned int	numberOfSurrogateSignalsIn,
		std::shared_ptr< Transformation >	transformationIn
	)

Constructor.

Parameters

numberOfSurrogateSignalsIn	The number of surrogate signals for this correspondence model.
transformationIn	The base transformation that is internally used to generate a transformation from input surrogate signals

~CorrespondenceModel()

CorrespondenceModel::~CorrespondenceModel

(

)

Destructor.

Member Function Documentation

ClearRecoveryModelParameters()

void CorrespondenceModel::ClearRecoveryModelParameters ( )

private

Free the memory allocated to store the recovery model parameters and set the corresponding pointer to null.

GetCorrespondenceModelAsImage()

std::vector< nifti_image * > CorrespondenceModel::GetCorrespondenceModelAsImage

(

)

Generates a vector with nifti images which can be saved. The data of the images will be allocated for the images, so they are detached from the internal representation of the correspondence model. The transformations can have multiple images so each image returned by this function reflects the number of images of the underlying transformation.

GetMaxParameterLength()

CorrespondenceModel::PrecisionType CorrespondenceModel::GetMaxParameterLength

(

PrecisionType *

parametersIn

)

Function used to calculate the gradient length, which in turn is used to normalise the gradient of the objective function. The correspondence model class defines the relationship between the respiratory correspondence model parameters and the transformation parameters. This is dependent on the number of surrogate signals. The maximum transformation parameter length will be calculated by the member transformation.

GetNumberOfParameters()

unsigned int CorrespondenceModel::GetNumberOfParameters ( )

inline

Get the number of parameters.

GetParameters()

CorrespondenceModel::PrecisionType * CorrespondenceModel::GetParameters

(

)

Get access to the current parameters. A pointer to \( \mathbf{R} \) is returned.

GetTransformation()

std::shared_ptr< Transformation > CorrespondenceModel::GetTransformation

(

)

Get a pointer to the transformation. Allow other objects, such as the ObjectiveFunction to use methods provided by the transformation

GetTransformationFromSurrogateSignal()

std::shared_ptr< Transformation > CorrespondenceModel::GetTransformationFromSurrogateSignal

(

const SurrogateSignalType &

surrogateSignalIn

)

Calculate the transformation parameters for a given surrogate signal on the basis of the current model parameters. \( \mathbf{M}_t(\mathbf{S}_t) \).

Parameters

surrogateSignalIn

Pointer to the surrogate values \( \mathbf{S}_t \). Get a transformatinon object corresponding to a surrogate signal. Checks the size of the surrogate signal.

GetTransformationParameterGradientWRTModelParameters()

void CorrespondenceModel::GetTransformationParameterGradientWRTModelParameters	(	CorrespondenceModel::PrecisionType *	transformationGradientIn,
		const SurrogateSignalType &	surrogateSignalIn,
		CorrespondenceModel::PrecisionType *	correspondenceModelGradientOut
	)

Get the gradient of the transformation parametres with respect to the model parametres. This function implements \( \frac{\partial \mathbf{M}_t }{\partial \mathbf{R} } \). Adds calculates the correspondence model gradient

InitialiseLevel()

void CorrespondenceModel::InitialiseLevel

(

unsigned int

levelIn

)

Initialise the given level. The transformation will be used to upsample the model parameters.

RecoverSavedModelParameters()

void CorrespondenceModel::RecoverSavedModelParameters

(

)

Recover correspondence model parameters that were previously saved. Parameters for recovery have to be saved previously using CorrespondenceModel::SaveCurrentModelParametersForRecovery().

SaveCurrentModelParametersForRecovery()

void CorrespondenceModel::SaveCurrentModelParametersForRecovery

(

)

Save the current model parameters to an internal variable such that they can be recovered if necessary by CorrespondenceModel::RecoverSavedModelParameters().

SetParameters() [1/2]

void CorrespondenceModel::SetParameters

(

const PrecisionType *

parametersIn

)

Set the current parameters. The internal data of \( \mathbf{R} \) will be updated (i.e. copied).

SetParameters() [2/2]

void CorrespondenceModel::SetParameters

(

const std::vector< nifti_image * > &

parameterImagesIn

)

Set the current parameters from the given input images. This function takes the image intensity and sets it as the correspondence model parameters. At the moment only the total number of parameters is tested.

Parameters

parameterImagesIn

Vectore with images holding the correspondence model parameters.

Member Data Documentation

modelParameters

PrecisionType* CorrespondenceModel::modelParameters

private

Pointer to the model parameters.

numberOfModelParameters

unsigned int CorrespondenceModel::numberOfModelParameters

private

The total number of parameters describing the correspondence model.

numberOfSurrogateSignals

unsigned int CorrespondenceModel::numberOfSurrogateSignals

private

The number of surrogate signals used.

numberOfTransformationParameters

unsigned int CorrespondenceModel::numberOfTransformationParameters

private

The number of parameters to describe the transformation completely.

recoveryModelParameters

PrecisionType* CorrespondenceModel::recoveryModelParameters

private

Pointer to model parameters that may be recovered.

transform

std::shared_ptr<Transformation> CorrespondenceModel::transform

private

The internal transformation object that is used to construct a transformation (copy) from a surrogate signal.

---

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

• motion-lib/CorrespondenceModel.h
• motion-lib/CorrespondenceModel.cpp