Course # January 10 March 20, Room GEN-S202 UC SF. Department of. 1/10/2012 N. Schuff 1/5 MI Square University of of California

Transcription

1 Medical Imaging Informatics Course # January 10 March 20, 2012 Tuesdays 9-11AM Room GEN-S202 1/5 MI Square 2012 Department of Veterans VA Medical Affairs Center & University of of California

2 Course Synopsis Overall Objective: Learn modern concepts of medical image processing and analysis Specific learning goals: Understand d principles i of imaging i informatics i and information theory Learn modern strategies of medical image processing Recognize new trends in image statistics 2/5 MI Square 2012

3 Course Outline Basics: This segment of the course will provide an overview of medical imaging modalities, image representations, foundations of imaging informatics and concepts of information theory, including definitions of entropy, complexity, and probability Image Processing Techniques: This segment will introduce general algorithms for rigid and non-rigid image registrations, the concept of atlas building, image segmentation, and new trends in computational anatomy. 3/5 MI Square 2012

4 Course Outline Statistical Analysis: This segment will present statistical concepts for image analysis, including linear modeling, univariate and multivariate analysis, independent components analysis, joint independent components analysis, as well as emerging trends in network analysis. Data Mining: This segment will introduce basic concepts of data mining for uni- and multimodal imaging applications. Different mining algorithms will be discussed, including decision trees, neural nets, Bayesian classifiers, support vector machines, nonparametric and reinforcement learning. 4/5 MI Square 2012

5 Timeline Dates /10 1/17 Lecture Titles Basics 1. Introduction to imaging informatics 2. Uses of information theory in imaging Presenters N. Schuff N. Schuff 1/24 1/31 2/07 2/14 Image Processing 3. Segmentation N. Schuff 4. Rigid Image Registration Cardenas/Tosun 5. Non-rigid Registration/Atlas building Cardenas/Tosun 6. Computational Anatomy Tosun 2/21 2/28 3/06 3/13 3/20 Statistical ti ti Analysis 7. Linear Modeling in Medical Imaging 8. Multimodal Imaging Analysis 9. Network Analysis Data Mining 10. Principles of Data Mining 11. Multimodal Image Mining J. Kornak D. Tosun V. Cardenas K. Young K. Young 5/5 MI Square 2012

6 WebLink 6/5 MI Square 2012

7 Contact Information Instructors Phone Norbert Schuff x4904 Karl Young Valerie Cardenas-Nicolson John Kornak Duygu Tosun 7/5 MI Square 2012

8 Bibliography 8/5 MI Square 2012

9 Foundation of Imaging (MRI) Foundations of medical imaging and signal recording -- Spectral transformations -- Information theory and principal component analysis -- Independent component analysis and blind source separation -- Dependent component analysis -- Pattern recognition techniques -- Fuzzy clustering and genetic algorithms -- Exploratory data analysis methods for fmri -- Low-frequency functional connectivity in fmri -- Classification of dynamic breast MR image data -- Dynamic cerebral contrast-enhanced perfusion MRI -- Skin lesion classification -- Microscopic slice image processing and automatic labeling -- NMR water artifact removal. 9/5 MI Square 2012

10 Foundation of Functional Imaging Department of 10/5 Radiology & MI Biomedical Square 2012 Imaging

11 Information Theory Part I: Theories of Perception and Learning Chapter 1: Bayesian Modelling of Visual Perception, by P. Mamassian, M. Landy and L. Maloney Chapter 2: Vision, Psychophysics, and Bayes, by P. Schrater and D. Kersten Chapter 3: Visual Cue Integration for Depth Perception, by R. Jacobs Chapter 4: Velocity Likelihoods in Biological and Machine Vision, by Y. Weiss and D. Fleet Chapter 5: Learning Motion Analysis, by W. Freeman, J. Haddon and E. Pasztor Chapter 6: Information Theoretic Approach to Neural Coding and Parameter Estimation: A Perspective, by J.-P. Nadal Chapter 7: From Generic to Specific: An Information Theoretic Perspective on the Value of High-Level Information,, by A. Yuille and J. Coughlan Chapter 8: Sparse Correlation Kernel Reconstruction and Superresolution, by C. Papageorgiou, F. Girosi and T. Poggio Department of 11/5 Radiology & MI Biomedical Square 2012 Imaging

12 Statistical Modeling Department of 12/5 Radiology & MI Biomedical Square 2012 Imaging

13 Statistical Modeling in Imaging Department of 13/5 Radiology & MI Biomedical Square 2012 Imaging 2 Probability Distributions BinaryVariables Thebetadistribution MultinomialVariables TheDirichletdistribution TheGaussianDistribution Conditional Gaussian distributions MarginalGaussiandistributions Bayes theorem for Gaussian variables Maximum likelihood for the Gaussian Sequentialestimation Bayesian inference for the Gaussian Student st-distribution Periodic variables MixturesofGaussians The Exponential Family Maximum likelihood and sufficient statistics Conjugatepriors Noninformativepriors Nonparametric Methods Kerneldensityestimators Nearest-neighbour methods Exercises Linear Models for Regression Linear Basis Function Models Maximum likelihood and least squares Geometry of least squares Sequentiallearning Regularized least squares Multipleoutputs TheBias-VarianceDecomposition BayesianLinearRegression Parameter distribution Predictivedistribution Equivalentkernel BayesianModelComparison TheEvidenceApproximation...165

14 Statistical Modeling Estimation Theory Basic concepts Properties of estimators Method of moments Least-squares estimation Linear least-squares method Nonlinear and generalized least squares * Maximum likelihood method Bayesian estimation * Minimum mean-square error estimator Wiener filtering Maximum a posteriori (MAP) estimator Concluding remarks and references 99 Problems Information Theory Entropy Definition of entropy Entropy and coding length Differential entropy Entropy of a transformation Mutual information Definition using entropy Definition using Kullback-Leibler divergence 110 Medical Imaging Informatics 2009, Department of 14/5 Radiology & MI Biomedical Square 2012 Imaging

15 R Scripts Introduction Computational Statistics and Statistical Computing The R Environment Getting Started with R Using the R Online Help System Functions Arrays, Data Frames, and Lists Workspace and Files Using Scripts Using Packages Graphics Probability and Statistics Review Random Variables and Probability Some Discrete Distributions Some Continuous Distributions Multivariate Normal Distribution Limit Theorems Statistics Bayes' Theorem and Bayesian Statistics Markov Chains Methods for Generating Random Variables Introduction And much more.. 15/5 MI Square 2012