itk::MattesMutualInformationImageToImageMetric< TFixedImage, TMovingImage > Class Template Reference
[Similarity Metrics of Registration Methods, Thread Unsafe classes, Similarity Metrics of Registration Methods]

#include <itkMattesMutualInformationImageToImageMetric.h>

Inheritance diagram for itk::MattesMutualInformationImageToImageMetric< TFixedImage, TMovingImage >:

[legend]Collaboration diagram for itk::MattesMutualInformationImageToImageMetric< TFixedImage, TMovingImage >:

[legend]List of all members.

Detailed Description

template<class TFixedImage, class TMovingImage>
class itk::MattesMutualInformationImageToImageMetric< TFixedImage, TMovingImage >

Computes the mutual information between two images to be registered using the method of Mattes et al.

MattesMutualInformationImageToImageMetric computes the mutual information between a fixed and moving image to be registered.

This class is templated over the FixedImage type and the MovingImage type.

The fixed and moving images are set via methods SetFixedImage() and SetMovingImage(). This metric makes use of user specified Transform and Interpolator. The Transform is used to map points from the fixed image to the moving image domain. The Interpolator is used to evaluate the image intensity at user specified geometric points in the moving image. The Transform and Interpolator are set via methods SetTransform() and SetInterpolator().

If a BSplineInterpolationFunction is used, this class obtain image derivatives from the BSpline interpolator. Otherwise, image derivatives are computed using central differencing.

Warning:: This metric assumes that the moving image has already been connected to the interpolator outside of this class.

The method GetValue() computes of the mutual information while method GetValueAndDerivative() computes both the mutual information and its derivatives with respect to the transform parameters.

The calculations are based on the method of Mattes et al [1,2] where the probability density distribution are estimated using Parzen histograms. Since the fixed image PDF does not contribute to the derivatives, it does not need to be smooth. Hence, a zero order (box car) BSpline kernel is used for the fixed image intensity PDF. On the other hand, to ensure smoothness a third order BSpline kernel is used for the moving image intensity PDF.

On Initialize(), the FixedImage is uniformly sampled within the FixedImageRegion. The number of samples used can be set via SetNumberOfSpatialSamples(). Typically, the number of spatial samples used should increase with the image size.

The option UseAllPixelOn() disables the random sampling and uses all the pixels of the FixedImageRegion in order to estimate the joint intensity PDF.

During each call of GetValue(), GetDerivatives(), GetValueAndDerivatives(), marginal and joint intensity PDF's values are estimated at discrete position or bins. The number of bins used can be set via SetNumberOfHistogramBins(). To handle data with arbitray magnitude and dynamic range, the image intensity is scale such that any contribution to the histogram will fall into a valid bin.

One the PDF's have been contructed, the mutual information is obtained by doubling summing over the discrete PDF values.

Notes: 1. This class returns the negative mutual information value. 2. This class in not thread safe due the private data structures used to the store the sampled points and the marginal and joint pdfs.

References: [1] "Nonrigid multimodality image registration" D. Mattes, D. R. Haynor, H. Vesselle, T. Lewellen and W. Eubank Medical Imaging 2001: Image Processing, 2001, pp. 1609-1620. [2] "PET-CT Image Registration in the Chest Using Free-form Deformations" D. Mattes, D. R. Haynor, H. Vesselle, T. Lewellen and W. Eubank IEEE Transactions in Medical Imaging. Vol.22, No.1, January 2003. pp.120-128. [3] "Optimization of Mutual Information for MultiResolution Image Registration" P. Thevenaz and M. Unser IEEE Transactions in Image Processing, 9(12) December 2000.