Transformation Class Reference

#include <Transformation.h>

Inheritance diagram for Transformation:

Public Types
typedef float	PrecisionType

Public Member Functions
virtual	~Transformation ()
virtual void	InitialiseLevel (unsigned int level)=0
virtual PrecisionType *	GetCopyOfParameters ()=0
virtual unsigned int	GetNumberOfParameters ()
virtual void	SetParameters (PrecisionType *paramsIn, bool parametersAreDisplacements)=0
virtual nifti_image *	GetDeformationVectorField (nifti_image *targetImageIn)=0
virtual PrecisionType *	GetDVFGradientWRTTransformationParameters (nifti_image *denseDVFIn)=0
virtual PrecisionType *	GetConstraintGradientWRTTransformationParameters ()=0
virtual double	GetConstraintValue ()=0
void	TransformImage (nifti_image *sourceImage, nifti_image *warpedImage)
void	TransformImageAdjoint (nifti_image *sourceImage, nifti_image *sourceWeightsImage, nifti_image *warpedImage, nifti_image *warpedWeightsImage)
virtual std::shared_ptr< Transformation >	DeepCopy ()=0
virtual void	DisplayTransformationParameters ()=0
virtual PrecisionType	GetSumOfPenaltyWeights ()=0
virtual void	GetImageGradientWRTDVF (nifti_image *sourceImage, nifti_image *outWarpedGradientImage)
void	ReorientateVectorImage (nifti_image *vectorFieldToReorientate, mat44 reorientationMatrix)
virtual PrecisionType	GetMaxTransformationParameterLength (PrecisionType *parametersIn)=0
virtual std::vector< nifti_image * >	GetTransformationAsImage ()=0
PrecisionType	GetPaddingValue ()

Protected Member Functions
bool	CheckDVFImageUpdateRequired (nifti_image *targetImageIn)
bool	CheckImageGeometryEquality (nifti_image *img1, nifti_image *img2)
virtual void	ClearDeformationVectorFieldImage ()
void	CubicSplineTransformImage3D (nifti_image *sourceImage, nifti_image *warpedImage)
void	CubicSplineTransformImage2D (nifti_image *sourceImage, nifti_image *warpedImage)
void	NearestNeighbourTransformImage3D (nifti_image *sourceImage, nifti_image *warpedImage)
void	NearestNeighbourTransformImage2D (nifti_image *sourceImage, nifti_image *warpedImage)
void	TrilinearTransformImage (nifti_image *sourceImage, nifti_image *warpedImage)
void	BilinearTransformImage (nifti_image *sourceImage, nifti_image *warpedImage)
void	CubicSplineTransformImageAdjoint3D (nifti_image *sourceImage, nifti_image *sourceWeightsImage, nifti_image *warpedImage, nifti_image *warpedWeightsImage)
void	CubicSplineTransformImageAdjoint2D (nifti_image *sourceImage, nifti_image *sourceWeightsImage, nifti_image *warpedImage, nifti_image *warpedWeightsImage)
void	NearestNeighbourTransformImageAdjoint3D (nifti_image *sourceImage, nifti_image *sourceWeightsImage, nifti_image *warpedImage, nifti_image *warpedWeightsImage)
void	NearestNeighbourTransformImageAdjoint2D (nifti_image *sourceImage, nifti_image *sourceWeightsImage, nifti_image *warpedImage, nifti_image *warpedWeightsImage)
void	TrilinearTransformImageAdjoint (nifti_image *sourceImage, nifti_image *sourceWeightsImage, nifti_image *warpedImage, nifti_image *warpedWeightsImage)
void	BilinearTransformImageAdjoint (nifti_image *sourceImage, nifti_image *sourceWeightsImage, nifti_image *warpedImage, nifti_image *warpedWeightsImage)
void	TrilinearImageGradient (nifti_image *sourceImage, nifti_image *resultGradientImage)
void	BilinearImageGradient (nifti_image *sourceImage, nifti_image *resultGradientImage)
void	CubicSplineImageGradient3D (nifti_image *sourceImage, nifti_image *resultGradientImage)
void	CubicSplineImageGradient2D (nifti_image *sourceImage, nifti_image *resultGradientImage)

Protected Attributes
bool	dvfImageUpdateRequired
	Indicates if the DVF requires re-calculation when it is required. More...
unsigned int	numberOfParameters
	The total number of parameters needed to describe the transformation. More...
int	interpolation
	Type of interpolation used. More...
PrecisionType	warpedPaddingValue
	Padding value for the warped image. More...
nifti_image *	deformationVectorFieldImage
	The deformation vector field image. More...

Detailed Description

Transformation class Abstract base class that must be reimplemented to obtain a functioning transformation class.

Member Typedef Documentation

PrecisionType

typedef float Transformation::PrecisionType

Constructor & Destructor Documentation

~Transformation()

virtual Transformation::~Transformation ( )

inlinevirtual

Member Function Documentation

BilinearImageGradient()

void Transformation::BilinearImageGradient ( nifti_image *  sourceImage, nifti_image *  resultGradientImage  )

protected

BilinearTransformImage()

void Transformation::BilinearTransformImage ( nifti_image *  sourceImage, nifti_image *  warpedImage  )

protected

BilinearTransformImageAdjoint()

void Transformation::BilinearTransformImageAdjoint ( nifti_image *  sourceImage, nifti_image *  sourceWeightsImage, nifti_image *  warpedImage, nifti_image *  warpedWeightsImage  )

protected

Todo:

: Decide if original switch statement is necessary at all.

Todo:

Todo:

: Remove reg_round macro with inline function

CheckDVFImageUpdateRequired()

bool Transformation::CheckDVFImageUpdateRequired ( nifti_image *  targetImageIn )

protected

Function to check if the DVF needs to be updated according to the target image

Parameters

targetImageIn

The geometry if the DVF is checked against the target image

CheckImageGeometryEquality()

bool Transformation::CheckImageGeometryEquality ( nifti_image *  img1, nifti_image *  img2  )

protected

Function that checks the basic geometry of two images. Is used by CheckDVFImageUpdateRequired() and checks the image size (nx, ny, and nz) as well as the qform/sform matrix depending on the qform/sform code. Note: Differences in higher dimensions (nu, nv, nt) are allowed to facilitd comparison between target image and DVF.

ClearDeformationVectorFieldImage()

void Transformation::ClearDeformationVectorFieldImage ( )

protectedvirtual

Helper function to delete the DVF image

CubicSplineImageGradient2D()

void Transformation::CubicSplineImageGradient2D ( nifti_image *  sourceImage, nifti_image *  resultGradientImage  )

protected

CubicSplineImageGradient3D()

void Transformation::CubicSplineImageGradient3D ( nifti_image *  sourceImage, nifti_image *  resultGradientImage  )

protected

CubicSplineTransformImage2D()

void Transformation::CubicSplineTransformImage2D ( nifti_image *  sourceImage, nifti_image *  warpedImage  )

protected

CubicSplineTransformImage3D()

void Transformation::CubicSplineTransformImage3D ( nifti_image *  sourceImage, nifti_image *  warpedImage  )

protected

CubicSplineTransformImageAdjoint2D()

void Transformation::CubicSplineTransformImageAdjoint2D ( nifti_image *  sourceImage, nifti_image *  sourceWeightsImage, nifti_image *  warpedImage, nifti_image *  warpedWeightsImage  )

protected

Todo:

: Decide if original switch statement is necessary at all.

CubicSplineTransformImageAdjoint3D()

void Transformation::CubicSplineTransformImageAdjoint3D ( nifti_image *  sourceImage, nifti_image *  sourceWeightsImage, nifti_image *  warpedImage, nifti_image *  warpedWeightsImage  )

protected

Todo:

: Decide if original switch statement is necessary at all.

DeepCopy()

virtual std::shared_ptr< Transformation > Transformation::DeepCopy ( )

pure virtual

Generate a deep copy of the transformation and return it.

Implemented in BSplineTransformation, and SlidingTransformation.

DisplayTransformationParameters()

virtual void Transformation::DisplayTransformationParameters ( )

pure virtual

Print the current transformation parameters to the command line.

Implemented in BSplineTransformation, and SlidingTransformation.

GetConstraintGradientWRTTransformationParameters()

virtual PrecisionType * Transformation::GetConstraintGradientWRTTransformationParameters ( )

pure virtual

Get the gradient of the regularisation/constraint with respect to the transformation parameters \( \frac{\partial \mathcal{R}_t}{\partial\mathbf{M}_t} \)

Implemented in BSplineTransformation, and SlidingTransformation.

GetConstraintValue()

virtual double Transformation::GetConstraintValue ( )

pure virtual

Get the value of the constraint term

Implemented in BSplineTransformation, and SlidingTransformation.

GetCopyOfParameters()

virtual PrecisionType * Transformation::GetCopyOfParameters ( )

pure virtual

Get a copy of the parameters. The memory will be allocated with malloc and filled by the derived class.

Implemented in BSplineTransformation, and SlidingTransformation.

GetDeformationVectorField()

virtual nifti_image * Transformation::GetDeformationVectorField ( nifti_image *  targetImageIn )

pure virtual

Get the deformation vector field according to the current parameters. This function is expected to return a pointer to the member variable of the DVF.

Parameters

targetImageIn

Nifti image is used to allocate the DVF. The DVF will remain part of the transformation class and will be deleted if the transformation is deleted.

Implemented in BSplineTransformation, and SlidingTransformation.

GetDVFGradientWRTTransformationParameters()

virtual PrecisionType * Transformation::GetDVFGradientWRTTransformationParameters ( nifti_image *  denseDVFIn )

pure virtual

Compute the gradient of the DVF with respect to the transformation parameters \( \frac{\partial \textbf{DVF}_t}{\partial \mathbf{M_t}} \).

Parameters

denseDVFIn

pointer to the nifti image for which the best match of transformation parameters will be found

Returns

pointer to the internally allocated output data of length numberOfParameters. Needs to be freed outside of this class since no internal references will be kept.

Implemented in BSplineTransformation, and SlidingTransformation.

GetImageGradientWRTDVF()

void Transformation::GetImageGradientWRTDVF ( nifti_image *  sourceImage, nifti_image *  outWarpedGradientImage  )

virtual

Calculate the gradient of an image with respect to the deformation vector field. This calculates \( \frac{\partial \mathbf{I}_{T_t}}{\partial \textbf{DVF}_t} \) and is not dependent on any parametrisation of a derived transformation.

Parameters

sourceImage	Nifti-image structure with the image from which the derivative will be calculated.
outWarpedGradientImage	Nifti image structure wich will be filled with the calculated gradient. Has to be allocated outside of this function.

GetMaxTransformationParameterLength()

virtual PrecisionType Transformation::GetMaxTransformationParameterLength ( PrecisionType *  parametersIn )

pure virtual

Function to determine the maximum length of a given set of parameters. This length depends on how the transformation parameters correspond internally to each other, i.e. how the x, y, and z component are stored.

Implemented in BSplineTransformation, and SlidingTransformation.

GetNumberOfParameters()

virtual unsigned int Transformation::GetNumberOfParameters ( )

inlinevirtual

Get the number of parameters required to describe this transformation completely.

GetPaddingValue()

PrecisionType Transformation::GetPaddingValue ( )

inline

Get the padding value.

GetSumOfPenaltyWeights()

virtual PrecisionType Transformation::GetSumOfPenaltyWeights ( )

pure virtual

Calculate the sum of the penalty weights

Implemented in BSplineTransformation, and SlidingTransformation.

GetTransformationAsImage()

virtual std::vector< nifti_image * > Transformation::GetTransformationAsImage ( )

pure virtual

Function that returns the current transformation a vecotr of images. These are not the DVFs, but images of the parameters instead. Hence it can only be implemented by any derived classes.

Implemented in BSplineTransformation, and SlidingTransformation.

InitialiseLevel()

virtual void Transformation::InitialiseLevel ( unsigned int  level )

pure virtual

Initialise the transformation and all internal parameters.

Parameters

level

The level which is to be initialised.

Implemented in BSplineTransformation, and SlidingTransformation.

NearestNeighbourTransformImage2D()

void Transformation::NearestNeighbourTransformImage2D ( nifti_image *  sourceImage, nifti_image *  warpedImage  )

protected

NearestNeighbourTransformImage3D()

void Transformation::NearestNeighbourTransformImage3D ( nifti_image *  sourceImage, nifti_image *  warpedImage  )

protected

NearestNeighbourTransformImageAdjoint2D()

void Transformation::NearestNeighbourTransformImageAdjoint2D ( nifti_image *  sourceImage, nifti_image *  sourceWeightsImage, nifti_image *  warpedImage, nifti_image *  warpedWeightsImage  )

protected

NearestNeighbourTransformImageAdjoint3D()

void Transformation::NearestNeighbourTransformImageAdjoint3D ( nifti_image *  sourceImage, nifti_image *  sourceWeightsImage, nifti_image *  warpedImage, nifti_image *  warpedWeightsImage  )

protected

ReorientateVectorImage()

void Transformation::ReorientateVectorImage	(	nifti_image *	vectorFieldToReorientate,
		mat44	reorientationMatrix
	)

Function to reorientate vectors in 5D nifti image. Vectors which are stored along the 5th (u) dimension will be reoriented

Parameters

vectorFieldToReorientate	Vector image to be reorientated
reorientationMatrix	4 x 4 Matrix used to reorientate vectors

SetParameters()

virtual void Transformation::SetParameters ( PrecisionType *  paramsIn, bool  parametersAreDisplacements  )

pure virtual

Set the transformation parameters

Implemented in BSplineTransformation, and SlidingTransformation.

TransformImage()

void Transformation::TransformImage	(	nifti_image *	sourceImage,
		nifti_image *	warpedImage
	)

Transform/warp an image; The warped image defines the geometry and will contain the resampled image contents.

Parameters

sourceImage	The source image will be resampeled into the warped image space according to the transformation parameters
warpedImage	The warped image defines the geometry of the internally calculated DVF and will also hold the actually warped image.

TransformImageAdjoint()

void Transformation::TransformImageAdjoint	(	nifti_image *	sourceImage,
		nifti_image *	sourceWeightsImage,
		nifti_image *	warpedImage,
		nifti_image *	warpedWeightsImage
	)

Transform an image using the "push-interpolation". This function was originally implemented specifically for reg_resp, hence the functionality will be provided by protected function of this class.

Parameters

sourceImage	Image that will be push-transformed into the warped image
sourceWeightsImage	Weights associated with the source image. This considers the image acquisition process where the acquired image may not have contributed equally to the voxel in the source image.
warpedImage	Image into which the intensities will be pushed. Note that intensities will be accumulated.
warpedWeightsImage	Warped (and accumulated) weights of the source image.

TrilinearImageGradient()

void Transformation::TrilinearImageGradient ( nifti_image *  sourceImage, nifti_image *  resultGradientImage  )

protected

TrilinearTransformImage()

void Transformation::TrilinearTransformImage ( nifti_image *  sourceImage, nifti_image *  warpedImage  )

protected

TrilinearTransformImageAdjoint()

void Transformation::TrilinearTransformImageAdjoint ( nifti_image *  sourceImage, nifti_image *  sourceWeightsImage, nifti_image *  warpedImage, nifti_image *  warpedWeightsImage  )

protected

Todo:

: Decide if original switch statement is necessary at all.

Member Data Documentation

deformationVectorFieldImage

nifti_image* Transformation::deformationVectorFieldImage

protected

The deformation vector field image.

dvfImageUpdateRequired

bool Transformation::dvfImageUpdateRequired

protected

Indicates if the DVF requires re-calculation when it is required.

interpolation

int Transformation::interpolation

protected

Type of interpolation used.

Todo:

Change into enumeration.

numberOfParameters

unsigned int Transformation::numberOfParameters

protected

The total number of parameters needed to describe the transformation.

warpedPaddingValue

PrecisionType Transformation::warpedPaddingValue

protected

Padding value for the warped image.

---

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

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