ITK  4.6.0
Insight Segmentation and Registration Toolkit
Segmentation/GeodesicActiveContourShapePriorLevelSetImageFilter.cxx
/*=========================================================================
*
* Copyright Insight Software Consortium
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0.txt
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
*=========================================================================*/
// Software Guide : BeginLatex
//
// In medical imaging applications, the general shape, location and
// orientation of an anatomical structure of interest is typically
// known \emph{a priori}. This information can be used to aid the
// segmentation process especially when image contrast is low or
// when the object boundary is not distinct.
//
// In \cite{Leventon2000}, Leventon \emph{et al.} extended the
// geodesic active contours method with an additional shape-influenced term in
// the driving PDE. The \doxygen{GeodesicActiveContourShapePriorLevelSetFilter}
// is a generalization of Leventon's approach and its use is illustrated
// in the following example.
//
// To support shape-guidance, the generic level set
// equation (Eqn(~\ref{eqn:LevelSetEquation})) is extended to incorporate a
// shape guidance term:
//
// \begin{equation}
// \label{eqn:ShapeInfluenceTerm}
// \xi \left(\psi^{*}(\mathbf{x}) - \psi(\mathbf{x})\right)
// \end{equation}
//
// where $\psi^{*}$ is the signed distance function of the ``best-fit'' shape
// with respect to a shape model. The new term has the effect of driving the
// contour towards the best-fit shape. The scalar $\xi$ weights the influence
// of the shape term in the overall evolution. In general, the best-fit shape
// is not known ahead of time and has to be iteratively estimated in
// conjunction with the contour evolution.
//
// As with the \doxygen{GeodesicActiveContourLevelSetImageFilter}, the
// GeodesicActiveContourShapePriorLevelSetImageFilter expects two input
// images: the first is an initial level set and the second a feature image
// that represents the image edge potential. The configuration of this
// example is quite similar to the example in
// Section~\ref{sec:GeodesicActiveContourImageFilter} and hence the description
// will focus on the new objects involved in the segmentation process as shown
// in Figure~\ref{fig:GeodesicActiveContourShapePriorCollaborationDiagram}.
//
// \begin{figure} \center
// \includegraphics[width=\textwidth]{GeodesicActiveContourShapePriorCollaborationDiagram}
// \itkcaption[GeodesicActiveContourShapePriorLevelSetImageFilter collaboration
// diagram]{Collaboration diagram for the GeodesicActiveContourShapePriorLevelSetImageFilter
// applied to a segmentation task.}
// \label{fig:GeodesicActiveContourShapePriorCollaborationDiagram}
// \end{figure}
//
// The process pipeline begins with centering the input image using the
// the \doxygen{ChangeInformationImageFilter} to simplify the estimation of the pose
// of the shape, to be explained later.
// The centered image is then smoothed using non-linear diffusion to
// remove noise and the gradient magnitude is computed from the smoothed image.
// For simplicity, this example uses the \doxygen{BoundedReciprocalImageFilter}
// to produce the edge potential image.
//
// The \doxygen{FastMarchingImageFilter} creates an initial level set using three
// user specified seed positions and a initial contour radius. Three seeds are
// used in this example to facilitate the segmentation of long narrow objects
// in a smaller number of iterations.
// The output of the FastMarchingImageFilter is passed
// as the input to the GeodesicActiveContourShapePriorLevelSetImageFilter.
// At then end of the segmentation process, the output level set is passed
// to the \doxygen{BinaryThresholdImageFilter} to produce a binary mask
// representing the segmented object.
//
// The remaining objects in
// Figure~\ref{fig:GeodesicActiveContourShapePriorCollaborationDiagram}
// are used for shape modeling and estimation.
// The \doxygen{PCAShapeSignedDistanceFunction} represents a statistical
// shape model defined by a mean signed distance and the first $K$
// principal components modes; while the \doxygen{Euler2DTransform} is used
// to represent the pose of the shape. In this implementation, the
// best-fit shape estimation problem is reformulated as a minimization problem
// where the \doxygen{ShapePriorMAPCostFunction} is the cost function to
// be optimized using the \doxygen{OnePlusOneEvolutionaryOptimizer}.
//
// It should be noted that, although particular shape model, transform
// cost function, and optimizer are used in this example, the implementation
// is generic, allowing different instances of these components to be
// plugged in. This flexibility allows a user to tailor the behavior of the
// segmentation process to suit the circumstances of the targeted application.
//
// Let's start the example by including the headers of the new filters
// involved in the segmentation.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// Next, we include the headers of the objects involved in shape
// modeling and estimation.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
#include "vnl/vnl_sample.h"
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// Given the numerous parameters involved in tuning this segmentation method
// it is not uncommon for a segmentation process to
// run for several minutes and still produce an unsatisfactory result. For debugging
// purposes it is quite helpful to track the evolution of the
// segmentation as it progresses. The following defines a
// custom \doxygen{Command} class
// for monitoring the RMS change and shape parameters at each iteration.
//
// \index{itk::Geodesic\-Active\-Contour\-Shape\-Prior\-LevelSet\-Image\-Filter!Monitoring}
// \index{itk::Shape\-Prior\-Segmentation\-Level\-Set\-Image\-Filter!Monitoring}
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
#include "itkCommand.h"
template<class TFilter>
class CommandIterationUpdate : public itk::Command
{
public:
typedef CommandIterationUpdate Self;
itkNewMacro( Self );
protected:
CommandIterationUpdate() {};
public:
void Execute(itk::Object *caller, const itk::EventObject & event)
{
Execute( (const itk::Object *) caller, event);
}
void Execute(const itk::Object * object, const itk::EventObject & event)
{
const TFilter * filter =
dynamic_cast< const TFilter * >( object );
if( typeid( event ) != typeid( itk::IterationEvent ) )
{ return; }
std::cout << filter->GetElapsedIterations() << ": ";
std::cout << filter->GetRMSChange() << " ";
std::cout << filter->GetCurrentParameters() << std::endl;
}
};
// Software Guide : EndCodeSnippet
int main( int argc, char *argv[] )
{
if( argc < 18 )
{
std::cerr << "Missing Parameters " << std::endl;
std::cerr << "Usage: " << argv[0];
std::cerr << " inputImage outputImage";
std::cerr << " seed1X seed1Y";
std::cerr << " seed2X seed2Y";
std::cerr << " seed3X seed3Y";
std::cerr << " initialDistance";
std::cerr << " sigma";
std::cerr << " propagationScaling shapePriorScaling";
std::cerr << " meanShapeImage numberOfModes shapeModeFilePattern";
std::cerr << " startX startY" << std::endl;
return 1;
}
// Software Guide : BeginLatex
//
// We define the image type using a particular pixel type and
// dimension. In this case we will use 2D \code{float} images.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
typedef float InternalPixelType;
const unsigned int Dimension = 2;
typedef itk::Image< InternalPixelType, Dimension > InternalImageType;
// Software Guide : EndCodeSnippet
// The following lines instantiate the thresholding filter that will
// process the final level set at the output of the
// GeodesicActiveContourLevelSetImageFilter.
//
typedef unsigned char OutputPixelType;
InternalImageType,
OutputImageType > ThresholdingFilterType;
ThresholdingFilterType::Pointer thresholder = ThresholdingFilterType::New();
thresholder->SetLowerThreshold( -1000.0 );
thresholder->SetUpperThreshold( 0.0 );
thresholder->SetOutsideValue( 0 );
thresholder->SetInsideValue( 255 );
// We instantiate reader and writer types in the following lines.
//
ReaderType::Pointer reader = ReaderType::New();
WriterType::Pointer writer = WriterType::New();
reader->SetFileName( argv[1] );
writer->SetFileName( argv[2] );
// The RescaleIntensityImageFilter type is declared below. This filter will
// renormalize image before sending them to writers.
//
InternalImageType,
OutputImageType > CastFilterType;
// The \doxygen{CurvatureAnisotropicDiffusionImageFilter} type is
// instantiated using the internal image type.
//
InternalImageType,
InternalImageType > SmoothingFilterType;
SmoothingFilterType::Pointer smoothing = SmoothingFilterType::New();
// The types of the
// GradientMagnitudeRecursiveGaussianImageFilter is
// instantiated using the internal image type.
//
InternalImageType,
InternalImageType > GradientFilterType;
GradientFilterType::Pointer gradientMagnitude = GradientFilterType::New();
// We declare now the type of the FastMarchingImageFilter that
// will be used to generate the initial level set in the form of a distance
// map.
//
InternalImageType,
InternalImageType > FastMarchingFilterType;
// Next we construct one filter of this class using the \code{New()}
// method.
//
FastMarchingFilterType::Pointer fastMarching = FastMarchingFilterType::New();
// Software Guide : BeginLatex
//
// The following line instantiate a
// \doxygen{GeodesicActiveContourShapePriorLevelSetImageFilter}
// using the \code{New()} method.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
InternalImageType, InternalImageType >
GeodesicActiveContourFilterType;
GeodesicActiveContourFilterType::Pointer geodesicActiveContour =
GeodesicActiveContourFilterType::New();
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// The \doxygen{ChangeInformationImageFilter} is the first filter in the preprocessing
// stage and is used to force the image origin to the center of the image.
//
// \index{itk::ChangeInformationImageFilter!CenterImageOn()}
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
InternalImageType > CenterFilterType;
CenterFilterType::Pointer center = CenterFilterType::New();
center->CenterImageOn();
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// In this example, we will use the bounded reciprocal $1/(1+x)$ of
// the image gradient magnitude as the edge potential feature image.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
InternalImageType,
InternalImageType > ReciprocalFilterType;
ReciprocalFilterType::Pointer reciprocal = ReciprocalFilterType::New();
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// In the GeodesicActiveContourShapePriorLevelSetImageFilter, scaling parameters
// are used to trade off between the propagation (inflation), the
// curvature (smoothing), the advection, and the shape influence terms.
// These parameters are set
// using methods \code{SetPropagationScaling()},
// \code{SetCurvatureScaling()}, \code{SetAdvectionScaling()} and
// \code{SetShapePriorScaling()}. In this
// example, we will set the curvature and advection scales to one and let
// the propagation and shape prior scale be command-line arguments.
//
// \index{itk::Geodesic\-Active\-Contour\-Shape\-Prior\-LevelSet\-Image\-Filter!SetPropagationScaling()}
// \index{itk::Shape\-Prior\-Segmentation\-Level\-Set\-Image\-Filter!SetPropagationScaling()}
// \index{itk::Geodesic\-Active\-Contour\-Shape\-Prior\-LevelSet\-Image\-Filter!SetCurvatureScaling()}
// \index{itk::Shape\-Prior\-Segmentation\-Level\-Set\-Image\-Filter!SetCurvatureScaling()}
// \index{itk::Geodesic\-Active\-Contour\-Shape\-Prior\-LevelSet\-Image\-Filter!SetAdvectionScaling()}
// \index{itk::Shape\-Prior\-Segmentation\-Level\-Set\-Image\-Filter!SetAdvectionScaling()}
//
// Software Guide : EndLatex
const double propagationScaling = atof( argv[11] );
const double shapePriorScaling = atof( argv[12] );
// Software Guide : BeginCodeSnippet
geodesicActiveContour->SetPropagationScaling( propagationScaling );
geodesicActiveContour->SetShapePriorScaling( shapePriorScaling );
geodesicActiveContour->SetCurvatureScaling( 1.0 );
geodesicActiveContour->SetAdvectionScaling( 1.0 );
// Software Guide : EndCodeSnippet
// Once activiated the level set evolution will stop if the convergence
// criteria or if the maximum number of iterations is reached. The
// convergence criteria is defined in terms of the root mean squared (RMS)
// change in the level set function. The evolution is said to have
// converged if the RMS change is below a user specified threshold. In a
// real application is desirable to couple the evolution of the zero set
// to a visualization module allowing the user to follow the evolution of
// the zero set. With this feedback, the user may decide when to stop the
// algorithm before the zero set leaks through the regions of low gradient
// in the contour of the anatomical structure to be segmented.
geodesicActiveContour->SetMaximumRMSError( 0.005 );
geodesicActiveContour->SetNumberOfIterations( 400 );
// Software Guide : BeginLatex
//
// Each iteration, the current ``best-fit'' shape is estimated from the
// edge potential image and the current contour. To increase speed, only
// information within the sparse field layers of the current contour is used
// in the estimation. The default number of sparse field layers is
// the same as
// the ImageDimension which does not contain enough information to get
// a reliable best-fit shape estimate. Thus, we override the default and
// set the number of layers to 4.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
geodesicActiveContour->SetNumberOfLayers( 4 );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// The filters are then connected in a pipeline as illustrated in
// Figure~\ref{fig:GeodesicActiveContourShapePriorCollaborationDiagram}.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
center->SetInput( reader->GetOutput() );
smoothing->SetInput( center->GetOutput() );
gradientMagnitude->SetInput( smoothing->GetOutput() );
reciprocal->SetInput( gradientMagnitude->GetOutput() );
geodesicActiveContour->SetInput( fastMarching->GetOutput() );
geodesicActiveContour->SetFeatureImage( reciprocal->GetOutput() );
thresholder->SetInput( geodesicActiveContour->GetOutput() );
writer->SetInput( thresholder->GetOutput() );
// Software Guide : EndCodeSnippet
// The CurvatureAnisotropicDiffusionImageFilter requires a couple of
// parameter to be defined. The following are typical values for $2D$
// images. However they may have to be adjusted depending on the amount of
// noise present in the input image. This filter has been discussed in
// section~\ref{sec:GradientAnisotropicDiffusionImageFilter}.
smoothing->SetTimeStep( 0.125 );
smoothing->SetNumberOfIterations( 5 );
smoothing->SetConductanceParameter( 9.0 );
// The GradientMagnitudeRecursiveGaussianImageFilter performs the
// equivalent of a convolution with a Gaussian kernel, followed by a
// derivative operator. The sigma of this Gaussian can be used to control
// the range of influence of the image edges. This filter has been discussed
// in Section~\ref{sec:GradientMagnitudeRecursiveGaussianImageFilter}
const double sigma = atof( argv[10] );
gradientMagnitude->SetSigma( sigma );
// The FastMarchingImageFilter requires the user to provide a seed
// point from which the level set will be generated. The user can actually
// pass not only one seed point but a set of them. Note that the
// FastMarchingImageFilter is used here only as a helper in the
// determination of an initial level set. We could have used the
// \doxygen{DanielssonDistanceMapImageFilter} in the same way.
//
// The seeds are passed stored in a container. The type of this
// container is defined as \code{NodeContainer} among the
// FastMarchingImageFilter traits.
//
typedef FastMarchingFilterType::NodeContainer NodeContainer;
typedef FastMarchingFilterType::NodeType NodeType;
NodeContainer::Pointer seeds = NodeContainer::New();
InternalImageType::IndexType seedPosition;
seedPosition[0] = atoi( argv[3] );
seedPosition[1] = atoi( argv[4] );
// Nodes are created as stack variables and initialized with a value and an
// \doxygen{Index} position. Note that here we assign the value of minus the
// user-provided distance to the unique node of the seeds passed to the
// FastMarchingImageFilter. In this way, the value will increment
// as the front is propagated, until it reaches the zero value corresponding
// to the contour. After this, the front will continue propagating until it
// fills up the entire image. The initial distance is taken here from the
// command line arguments. The rule of thumb for the user is to select this
// value as the distance from the seed points at which she want the initial
// contour to be.
const double initialDistance = atof( argv[9] );
NodeType node;
const double seedValue = - initialDistance;
node.SetValue( seedValue );
node.SetIndex( seedPosition );
// The list of nodes is initialized and then every node is inserted using
// the \code{InsertElement()}.
seeds->Initialize();
seeds->InsertElement( 0, node );
seedPosition[0] = atoi( argv[5] );
seedPosition[1] = atoi( argv[6] );
node.SetIndex( seedPosition );
seeds->InsertElement( 1, node );
seedPosition[0] = atoi( argv[7] );
seedPosition[1] = atoi( argv[8] );
node.SetIndex( seedPosition );
seeds->InsertElement( 2, node );
// The set of seed nodes is passed now to the
// FastMarchingImageFilter with the method
// \code{SetTrialPoints()}.
//
fastMarching->SetTrialPoints( seeds );
// Since the FastMarchingImageFilter is used here just as a
// Distance Map generator. It does not require a speed image as input.
// Instead the constant value $1.0$ is passed using the
// \code{SetSpeedConstant()} method.
//
fastMarching->SetSpeedConstant( 1.0 );
// Here we configure all the writers required to see the intermediate
// outputs of the pipeline. This is added here only for
// pedagogical/debugging purposes. These intermediate output are normaly not
// required. Only the output of the final thresholding filter should be
// relevant. Observing intermediate output is helpful in the process of
// fine tuning the parameters of filters in the pipeline.
//
CastFilterType::Pointer caster1 = CastFilterType::New();
CastFilterType::Pointer caster2 = CastFilterType::New();
CastFilterType::Pointer caster3 = CastFilterType::New();
CastFilterType::Pointer caster4 = CastFilterType::New();
WriterType::Pointer writer1 = WriterType::New();
WriterType::Pointer writer2 = WriterType::New();
WriterType::Pointer writer3 = WriterType::New();
WriterType::Pointer writer4 = WriterType::New();
caster1->SetInput( smoothing->GetOutput() );
writer1->SetInput( caster1->GetOutput() );
writer1->SetFileName("GeodesicActiveContourShapePriorImageFilterOutput1.png");
caster1->SetOutputMinimum( 0 );
caster1->SetOutputMaximum( 255 );
writer1->Update();
caster2->SetInput( gradientMagnitude->GetOutput() );
writer2->SetInput( caster2->GetOutput() );
writer2->SetFileName("GeodesicActiveContourShapePriorImageFilterOutput2.png");
caster2->SetOutputMinimum( 0 );
caster2->SetOutputMaximum( 255 );
writer2->Update();
caster3->SetInput( reciprocal->GetOutput() );
writer3->SetInput( caster3->GetOutput() );
writer3->SetFileName("GeodesicActiveContourShapePriorImageFilterOutput3.png");
caster3->SetOutputMinimum( 0 );
caster3->SetOutputMaximum( 255 );
writer3->Update();
caster4->SetInput( fastMarching->GetOutput() );
writer4->SetInput( caster4->GetOutput() );
writer4->SetFileName("GeodesicActiveContourShapePriorImageFilterOutput4.png");
caster4->SetOutputMinimum( 0 );
caster4->SetOutputMaximum( 255 );
// The FastMarchingImageFilter requires the user to specify the
// size of the image to be produced as output. This is done using the
// \code{SetOutputRegion()}. Note that the size is obtained here from the
// output image of the centering filter. The size of this image is valid
// only after the \code{Update()} methods of this filter has been called
// directly or indirectly.
//
fastMarching->SetOutputRegion(
center->GetOutput()->GetBufferedRegion() );
fastMarching->SetOutputSpacing(
center->GetOutput()->GetSpacing() );
fastMarching->SetOutputOrigin(
center->GetOutput()->GetOrigin() );
// Software Guide : BeginLatex
//
// Next, we define the shape model. In this example,
// we use an implicit shape model based on the principal components
// such that:
//
// \begin{equation}
// \psi^{*}(\mathbf{x}) = \mu(\mathbf{x}) + \sum_k \alpha_k u_k(\mathbf{x})
// \end{equation}
//
// where $\mu(\mathbf{x})$ is the mean signed distance computed from training
// set of segmented objects and $u_k(\mathbf{x})$ are the first $K$ principal
// components of the offset (signed distance - mean).
// The coefficients $\{\alpha_k\}$ form the
// set of \emph{shape} parameters.
//
// Given a set of training data, the \doxygen{ImagePCAShapeModelEstimator}
// can be used to obtain
// the mean and principal mode shape images required by PCAShapeSignedDistanceFunction.
//
// \index{itk::PCAShapeSignedDistanceFunction!New()}
// \index{itk::PCAShapeSignedDistanceFunction!SetNumberOfPrincipalComponents()}
//
//
// Software Guide : EndLatex
const unsigned int numberOfPCAModes = atoi( argv[14] );
// Software Guide : BeginCodeSnippet
double,
Dimension,
InternalImageType > ShapeFunctionType;
ShapeFunctionType::Pointer shape = ShapeFunctionType::New();
shape->SetNumberOfPrincipalComponents( numberOfPCAModes );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// In this example, we will read the mean shape and
// principal mode images from file. We will assume that
// the filenames of the mode images form a numeric series starting from index 0.
//
// \index{itk::PCAShapeSignedDistanceFunction!SetMeanImage()}
// \index{itk::PCAShapeSignedDistanceFunction!SetPrincipalComponentsImages()}
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
ReaderType::Pointer meanShapeReader = ReaderType::New();
meanShapeReader->SetFileName( argv[13] );
meanShapeReader->Update();
std::vector<InternalImageType::Pointer> shapeModeImages( numberOfPCAModes );
fileNamesCreator->SetStartIndex( 0 );
fileNamesCreator->SetEndIndex( numberOfPCAModes - 1 );
fileNamesCreator->SetSeriesFormat( argv[15] );
const std::vector<std::string> & shapeModeFileNames =
fileNamesCreator->GetFileNames();
for ( unsigned int k = 0; k < numberOfPCAModes; k++ )
{
ReaderType::Pointer shapeModeReader = ReaderType::New();
shapeModeReader->SetFileName( shapeModeFileNames[k].c_str() );
shapeModeReader->Update();
shapeModeImages[k] = shapeModeReader->GetOutput();
}
shape->SetMeanImage( meanShapeReader->GetOutput() );
shape->SetPrincipalComponentImages( shapeModeImages );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// Further we assume that the shape modes have been normalized
// by multiplying with the corresponding singular value. Hence,
// we can set the principal component standard deviations to all
// ones.
//
// \index{itk::PCAShapeSignedDistanceFunction!Set\-Principal\-Component\-Standard\-Deviations()}
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
ShapeFunctionType::ParametersType pcaStandardDeviations( numberOfPCAModes );
pcaStandardDeviations.Fill( 1.0 );
shape->SetPrincipalComponentStandardDeviations( pcaStandardDeviations );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// Next, we instantiate a \doxygen{Euler2DTransform} and connect it to the
// PCASignedDistanceFunction. The transform represent
// the pose of the shape. The parameters of the transform
// forms the set of \emph{pose} parameters.
//
// \index{itk::PCAShapeSignedDistanceFunction!SetTransform()}
// \index{itk::ShapeSignedDistanceFunction!SetTransform()}
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
typedef itk::Euler2DTransform<double> TransformType;
TransformType::Pointer transform = TransformType::New();
shape->SetTransform( transform );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// Before updating the level set at each iteration, the parameters
// of the current best-fit shape is estimated by minimizing the
// \doxygen{ShapePriorMAPCostFunction}. The cost function is composed of
// four terms: contour fit, image fit, shape prior and pose prior.
// The user can specify the weights applied to each term.
//
// \index{itk::ShapePriorMAPCostFunction!SetWeights()}
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
InternalImageType,
InternalPixelType > CostFunctionType;
CostFunctionType::Pointer costFunction = CostFunctionType::New();
CostFunctionType::WeightsType weights;
weights[0] = 1.0; // weight for contour fit term
weights[1] = 20.0; // weight for image fit term
weights[2] = 1.0; // weight for shape prior term
weights[3] = 1.0; // weight for pose prior term
costFunction->SetWeights( weights );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// Contour fit measures the likelihood of seeing the current
// evolving contour for a given set of shape/pose parameters.
// This is computed by counting the number of pixels inside
// the current contour but outside the current shape.
//
// Image fit measures the likelihood of seeing certain image
// features for a given set of shape/pose parameters. This is
// computed by assuming that ( 1 - edge potential ) approximates
// a zero-mean, unit variance Gaussian along the normal of
// the evolving contour. Image fit is then computed by computing
// the Laplacian goodness of fit of the Gaussian:
//
// \begin{equation}
// \sum \left( G(\psi(\mathbf{x})) - |1 - g(\mathbf{x})| \right)^2
// \end{equation}
//
// where $G$ is a zero-mean, unit variance Gaussian and $g$ is
// the edge potential feature image.
//
// The pose parameters are assumed to have a uniform distribution
// and hence do not contribute to the cost function.
// The shape parameters are assumed to have a Gaussian distribution.
// The parameters of the distribution are user-specified. Since we
// assumed the principal modes have already been normalized,
// we set the distribution to zero mean and unit variance.
//
// \index{itk::ShapePriorMAPCostFunction!SetShapeParameterMeans()}
// \index{itk::ShapePriorMAPCostFunction!SetShapeParameterStandardDeviations()}
//
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
CostFunctionType::ArrayType mean( shape->GetNumberOfShapeParameters() );
CostFunctionType::ArrayType stddev( shape->GetNumberOfShapeParameters() );
mean.Fill( 0.0 );
stddev.Fill( 1.0 );
costFunction->SetShapeParameterMeans( mean );
costFunction->SetShapeParameterStandardDeviations( stddev );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// In this example, we will use the \doxygen{OnePlusOneEvolutionaryOptimizer}
// to optimize the cost function.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
typedef itk::OnePlusOneEvolutionaryOptimizer OptimizerType;
OptimizerType::Pointer optimizer = OptimizerType::New();
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// The evolutionary optimization algorithm is based on testing
// random permutations of the parameters. As such, we need to provide
// the optimizer with a random number generator. In the following lines,
// we create a \doxygen{NormalVariateGenerator}, seed it, and
// connect it to the optimizer.
//
// \index{itk::Statistics::NormalVariateGenerator!Initialize()}
// \index{itk::OnePlusOneEvolutionaryOptimizer!SetNormalVariateGenerator()}
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
GeneratorType::Pointer generator = GeneratorType::New();
generator->Initialize( 20020702 );
optimizer->SetNormalVariateGenerator( generator );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// The cost function has $K+3$ parameters. The first $K$
// parameters are the principal component multipliers, followed
// by the 2D rotation parameter (in radians) and the x- and
// y- translation parameters (in mm). We need to carefully
// scale the different types of parameters to compensate
// for the differences in the dynamic ranges of the parameters.
//
// \index{itk::OnePlusOneEvolutionaryOptimizer!SetScales()}
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
OptimizerType::ScalesType scales( shape->GetNumberOfParameters() );
scales.Fill( 1.0 );
for( unsigned int k = 0; k < numberOfPCAModes; k++ )
{
scales[k] = 20.0; // scales for the pca mode multiplier
}
scales[numberOfPCAModes] = 350.0; // scale for 2D rotation
optimizer->SetScales( scales );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// Next, we specify the initial radius, the shrink and
// grow mutation factors and termination criteria of the optimizer.
// Since the best-fit shape is re-estimated each iteration of
// the curve evolution, we do not need to spend too much time finding the true
// minimizing solution each time; we only need to head towards it. As such,
// we only require a small number of optimizer iterations.
//
// \index{itk::OnePlusOneEvolutionaryOptimizer!Initialize()}
// \index{itk::OnePlusOneEvolutionaryOptimizer!SetEpsilon()}
// \index{itk::OnePlusOneEvolutionaryOptimizer!SetMaximumIteration()}
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
double initRadius = 1.05;
double grow = 1.1;
double shrink = pow(grow, -0.25);
optimizer->Initialize(initRadius, grow, shrink);
optimizer->SetEpsilon(1.0e-6); // minimal search radius
optimizer->SetMaximumIteration(15);
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// Before starting the segmentation process we need to also supply the initial
// best-fit shape estimate. In this example, we start with the unrotated mean shape
// with the initial x- and y- translation specified through command-line
// arguments.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
ShapeFunctionType::ParametersType parameters(
shape->GetNumberOfParameters() );
parameters.Fill( 0.0 );
parameters[numberOfPCAModes + 1] = atof( argv[16] ); // startX
parameters[numberOfPCAModes + 2] = atof( argv[17] ); // startY
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// Finally, we connect all the components to the filter and add our
// observer.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
geodesicActiveContour->SetShapeFunction( shape );
geodesicActiveContour->SetCostFunction( costFunction );
geodesicActiveContour->SetOptimizer( optimizer );
geodesicActiveContour->SetInitialParameters( parameters );
typedef CommandIterationUpdate<GeodesicActiveContourFilterType> CommandType;
CommandType::Pointer observer = CommandType::New();
geodesicActiveContour->AddObserver( itk::IterationEvent(), observer );
// Software Guide : EndCodeSnippet
// Software Guide : BeginLatex
//
// The invocation of the \code{Update()} method on the writer triggers the
// execution of the pipeline. As usual, the call is placed in a
// \code{try/catch} block to handle exceptions should errors occur.
//
// Software Guide : EndLatex
// Software Guide : BeginCodeSnippet
try
{
writer->Update();
}
catch( itk::ExceptionObject & excep )
{
std::cerr << "Exception caught !" << std::endl;
std::cerr << excep << std::endl;
}
// Software Guide : EndCodeSnippet
// Print out some useful information
std::cout << std::endl;
std::cout << "Max. no. iterations: " << geodesicActiveContour->GetNumberOfIterations() << std::endl;
std::cout << "Max. RMS error: " << geodesicActiveContour->GetMaximumRMSError() << std::endl;
std::cout << std::endl;
std::cout << "No. elpased iterations: " << geodesicActiveContour->GetElapsedIterations() << std::endl;
std::cout << "RMS change: " << geodesicActiveContour->GetRMSChange() << std::endl;
std::cout << "Parameters: " << geodesicActiveContour->GetCurrentParameters() << std::endl;
writer4->Update();
// The following writer type is used to save the output of the time-crossing
// map in a file with apropiate pixel representation. The advantage of saving
// this image in native format is that it can be used with a viewer to help
// determine an appropriate threshold to be used on the output of the
// fastmarching filter.
//
typedef itk::ImageFileWriter< InternalImageType > InternalWriterType;
InternalWriterType::Pointer mapWriter = InternalWriterType::New();
mapWriter->SetInput( fastMarching->GetOutput() );
mapWriter->SetFileName("GeodesicActiveContourShapePriorImageFilterOutput4.mha");
mapWriter->Update();
InternalWriterType::Pointer speedWriter = InternalWriterType::New();
speedWriter->SetInput( reciprocal->GetOutput() );
speedWriter->SetFileName("GeodesicActiveContourShapePriorImageFilterOutput3.mha");
speedWriter->Update();
InternalWriterType::Pointer gradientWriter = InternalWriterType::New();
gradientWriter->SetInput( gradientMagnitude->GetOutput() );
gradientWriter->SetFileName("GeodesicActiveContourShapePriorImageFilterOutput2.mha");
gradientWriter->Update();
// Also write out the initial and final best fit shape
ShapeFunctionType,
InternalImageType,
InternalImageType > EvaluatorFilterType;
EvaluatorFilterType::Pointer evaluator = EvaluatorFilterType::New();
evaluator->SetInput( geodesicActiveContour->GetOutput() );
evaluator->SetFunction( shape );
shape->SetParameters( geodesicActiveContour->GetInitialParameters() );
thresholder->SetInput( evaluator->GetOutput() );
writer->SetFileName( "GeodesicActiveContourShapePriorImageFilterOutput5.png" );
writer->Update();
shape->SetParameters( geodesicActiveContour->GetCurrentParameters() );
evaluator->Modified();
writer->SetFileName( "GeodesicActiveContourShapePriorImageFilterOutput6.png" );
writer->Update();
// Software Guide : BeginLatex
//
// Deviating from previous examples, we will demonstrate this example using
// \code{BrainMidSagittalSlice.png}
// (Figure~\ref{fig:GeodesicActiveContourShapePriorImageFilterOutput}, left)
// from the \code{Examples/Data} directory.
// The aim here is to segment the corpus callosum from the image using a shape model
// defined by \code{CorpusCallosumMeanShape.mha} and the first three principal
// components \code{CorpusCallosumMode0.mha}, \code{CorpusCallosumMode1.mha} and
// \code{CorpusCallosumMode12.mha}. As shown in Figure~\ref{fig:CorpusCallosumPCAModes},
// the first mode captures scaling, the second mode captures the shifting of mass between
// the rostrum and the splenium and the third mode captures the degree of curvature.
// Segmentation results with and without shape
// guidance are shown in
// Figure~\ref{fig:GeodesicActiveContourShapePriorImageFilterOutput2}.
//
//
// \begin{figure} \center
// \includegraphics[width=0.30\textwidth]{BrainMidSagittalSlice}
// \includegraphics[width=0.30\textwidth]{GeodesicActiveContourShapePriorImageFilterOutput5}
// \itkcaption[GeodesicActiveContourShapePriorImageFilter input image and initial model]{
// The input image to the GeodesicActiveContourShapePriorLevelSetImageFilter is a
// synthesized MR-T1 mid-sagittal slice ($217 \times 180$ pixels, $1 \times 1$ mm spacing)
// of the brain (left) and the initial best-fit shape
// (right) chosen to roughly overlap the corpus callosum in the image to be segmented.}
//
// \label{fig:GeodesicActiveContourShapePriorImageFilterOutput}
// \end{figure}
//
//
// \begin{figure}
// \center
// \begin{tabular}{cccc}
// & $-3\sigma$ & mean & $+3\sigma$ \\ mode 0: &
// \includegraphics[width=0.10\textwidth]{CorpusCallosumModeMinus0} &
// \includegraphics[width=0.10\textwidth]{CorpusCallosumMeanShape} &
// \includegraphics[width=0.10\textwidth]{CorpusCallosumModePlus0} \\ mode 1: &
// \includegraphics[width=0.10\textwidth]{CorpusCallosumModeMinus1} &
// \includegraphics[width=0.10\textwidth]{CorpusCallosumMeanShape} &
// \includegraphics[width=0.10\textwidth]{CorpusCallosumModePlus1} \\ mode 2: &
// \includegraphics[width=0.10\textwidth]{CorpusCallosumModeMinus2} &
// \includegraphics[width=0.10\textwidth]{CorpusCallosumMeanShape} &
// \includegraphics[width=0.10\textwidth]{CorpusCallosumModePlus2} \\ \end{tabular}
// \itkcaption[Corpus callosum PCA modes]{First three PCA modes of a low-resolution
// ($58 \times 31$ pixels, $2 \times 2$ mm spacing) corpus callosum model used in the
// shape guided geodesic active contours example.}
//
// \label{fig:CorpusCallosumPCAModes}
// \end{figure}
//
//
//
// A sigma value of $1.0$ was used to compute the image gradient and the
// propagation and shape prior scaling are respectively set to $0.5$ and $0.02$.
// An initial level set was created by placing one seed point in the
// rostrum $(60,102)$, one in the splenium $(120, 85)$ and one
// centrally in the body $(88,83)$ of the corpus callosum with
// an initial radius of $6$ pixels at each seed position.
// The best-fit shape was initially placed with a translation of
// $(10,0)$mm so that it roughly overlapped
// the corpus callosum in the image as shown in
// Figure~\ref{fig:GeodesicActiveContourShapePriorImageFilterOutput} (right).
//
//
// From Figure~\ref{fig:GeodesicActiveContourShapePriorImageFilterOutput2} it can be
// observed that without
// shape guidance (left), segmentation using geodesic active contour leaks in the
// regions where the corpus callosum blends into the surrounding brain tissues. With
// shape guidance (center), the segmentation is constrained by the global shape model
// to prevent leaking.
//
// The final best-fit shape parameters after the segmentation process is:
//
// \begin{verbatim}
// Parameters: [-0.384988, -0.578738, 0.557793, 0.275202, 16.9992, 4.73473]
// \end{verbatim}
//
// and is shown in
// Figure~\ref{fig:GeodesicActiveContourShapePriorImageFilterOutput2} (right). Note that a
// $0.28$ radian ($15.8$ degree) rotation has been introduced to match the model to
// the corpus callosum in the image. Additionally, a negative weight for the first
// mode shrinks the size relative to the mean shape. A negative weight for the second mode
// shifts the mass to splenium, and a positive weight for the third mode
// increases the curvature. It can also be observed that the final segmentation is
// a combination of the best-fit shape with additional local deformation. The combination
// of both global and local shape allows the segmentation to capture fine details not represented
// in the shape model.
//
//
// \begin{figure} \center
// \includegraphics[width=0.30\textwidth]{GeodesicActiveContourShapePriorImageFilterOutput1}
// \includegraphics[width=0.30\textwidth]{GeodesicActiveContourShapePriorImageFilterOutput2}
// \includegraphics[width=0.30\textwidth]{GeodesicActiveContourShapePriorImageFilterOutput6}
// \itkcaption[GeodesicActiveContourShapePriorImageFilter segmentations]{Corpus callosum
// segmentation using geodesic active contours without (left) and with (center) shape guidance.
// The image on the right represents the best-fit shape at the end of the segmentation process.}
//
// \label{fig:GeodesicActiveContourShapePriorImageFilterOutput2}
// \end{figure}
//
//
// Software Guide : EndLatex
return 0;
}