1st Edition

Stochastic Modeling for Medical Image Analysis

    304 Pages 188 Color Illustrations
    by CRC Press

    304 Pages 188 Color Illustrations
    by CRC Press

    Stochastic Modeling for Medical Image Analysis provides a brief introduction to medical imaging, stochastic modeling, and model-guided image analysis.

    Today, image-guided computer-assisted diagnostics (CAD) faces two basic challenging problems. The first is the computationally feasible and accurate modeling of images from different modalities to obtain clinically useful information. The second is the accurate and fast inferring of meaningful and clinically valid CAD decisions and/or predictions on the basis of model-guided image analysis.

    To help address this, this book details original stochastic appearance and shape models with computationally feasible and efficient learning techniques for improving the performance of object detection, segmentation, alignment, and analysis in a number of important CAD applications.

    The book demonstrates accurate descriptions of visual appearances and shapes of the goal objects and their background to help solve a number of important and challenging CAD problems. The models focus on the first-order marginals of pixel/voxel-wise signals and second- or higher-order Markov-Gibbs random fields of these signals and/or labels of regions supporting the goal objects in the lattice.

    This valuable resource presents the latest state of the art in stochastic modeling for medical image analysis while incorporating fully tested experimental results throughout.

    Medical Imaging Modalities
    Magnetic Resonance Imaging
    Computed Tomography
    Ultrasound Imaging
    Nuclear Medical Imaging (Nuclide Imaging)
    Bibliographic and Historical Notes

    From Images to Graphical Models

    Basics of Image Modeling
    Pixel/Voxel Interactions and Neighborhoods
    Exponential Families of Probability Distributions
    Appearance and Shape Modeling
    Bibliographic and Historical Notes

    IRF Models: Estimating Marginals

    Basic Independent Random Fields
    Supervised and Unsupervised Learning
    Expectation-Maximization to Identify Mixtures
    Gaussian Linear Combinations versus Mixtures
    Bibliographic and Historical Notes

    Markov-Gibbs Random Field Models: Estimating Signal Interactions
    Generic Kth-Order MGRFs
    Common Second- and Higher-Order MGRFs
    Learning Second-Order Interaction Structures
    Bibliographic and Historical Notes

    Applications: Image Alignment

    General Image Alignment Frameworks
    Global Alignment by Learning an Appearance Prior
    Bibliographic and Historical Notes

    Segmenting Multimodal Images

    Joint MGRF of Images and Region Maps
    Experimental Validation
    Bibliographic and Historical Notes
    Performance Evaluation and Validation

    Segmenting with Deformable Models

    Appearance-Based Segmentation
    Shape and Appearance-Based Segmentation
    Bibliographic and Historical Notes

    Segmenting with Shape and Appearance Priors

    Learning a Shape Prior
    Evolving a Deformable Boundary
    Experimental Validation
    Bibliographic and Historical Notes

    Cine Cardiac MRI Analysis

    Segmenting Myocardial Borders
    Wall Thickness Analysis
    Experimental Results
    Bibliographic and Historical Notes

    Sizing Cardiac Pathologies

    LV Wall Segmentation
    Identifying the Pathological Tissue
    Quantifying the Myocardial Viability
    Performance Evaluation and Validation
    Bibliographic and Historical Notes


    Ayman El-Baz, PhD, associate professor, Department of Bioengineering, University of Louisville, Kentucky, USA

    Georgy Gimel’farb, professor of computer science, University of Auckland, New Zealand

    Jasjit S. Suri, PhD, MBA, CEO, Global Biomedical Technologies, Inc., Roseville, California, USA