AHM 2006:ProjectsJointRegistrationSegmentation

Define Joint Registration and Segmentation Framework

We are interested in extending the ITK registration framework with our joint atlas registration and MR segmentation algorithm. Our work is described in detail in the progress report Shape based Segmentation and Registration. We have implemented the algorithm in VTK and are planning to release it into the 3D Slicer 2.6.

File:2006 Programming-JointRegistrationSegmentation.ppt

Status

• Initial code is in the NAMIC Sandbox under directory /JointRegistrationSegmentation that implements the registration part
  • http://www.na-mic.org/svn/NAMICSandBox/JointRegistrationSegmentation/
  • Using the customized Metric that Kilian developed http://www.na-mic.org/svn/NAMICSandBox/JointRegistrationSegmentation/itkKullbackLeiblerDivergenceImageToImageMetric.h

• Possiblity of taking advantage of the BayesianSegmentationModule in order to implement the Expectation Maximization
  • http://www.na-mic.org/svn/NAMICSandBox/BayesianSegmentationModule/

Pending Tasks

• Update the RecursiveGaussianImageFilter to work on VectorImages
  • http://www.itk.org/Insight/Doxygen/html/classitk_1_1GradientRecursiveGaussianImageFilter.html

• Create a connection between the output of the MRIBiasFieldCorrectionImageFilter and the itkBayesianClassifierInitializationImageFilter in order to recompute the new weight from the corrected intensity values of the input MRI image, and the new GaussianDensity functions
  • http://www.itk.org/Insight/Doxygen/html/classitk_1_1MRIBiasFieldCorrectionFilter.html