Image-Based Tissue Analysis and the Clinical Applications

This presentation briefly presents image-based tissue analysis methods and clinical applications. In the first part of this presentation, a three dimensional(3-D) tissue analysis method on MR images is discussed. First, the MR image volume is modeled by standard finite normal mixture(SFNM) distribution. Then the tissue quantification is achieved through (1) model selection by minimum description length (MDL) criterion; (2) parameter estimation by optimal histogram quantization and a fast EM algorithm using the global 3-D histogram rather than conventionally the raw data. Finally, a 3-D segmentation method is employed by the maximum likelihood (ML) classification and contextual Bayesian relaxation labeling(CBRL). The CBRL is developed to obtain a consistent labeling solution, based on localized SFNM formulation regularities. The method has been applied to partial volume correction for PET brain images and change analysis for MR breast images. The second part in this presentation is to conduct some preliminary studies for performing independent component analysis (ICA) on independent component imaging to identify and extract the independent components from a kinetic study. When the sources are mixtures of independent and dependent random variables, a possible solution is presented, which uses a feature selection step to extract independent variables from the sources in observation space.