[[Overview]] [[Location]] [[People]] [[Weekly Schedule]] [[Readings]] [[Projects]] [[Useful Links]] [[Announcements]]
@media screen\n{\n\nbody {\n font: 13px/125% "Lucida Grande", "Trebuchet MS", "Bitstream Vera Sans", Verdana, Helvetica, sans-serif;\n}\n\n}\n\n@media print\n{\n\nbody {\n font-size: 6.2pt;\n font-family: "Lucida Grande", "Bitstream Vera Sans", Helvetica, Verdana, Arial, sans-serif;\n}\n\n}
config.options.chkHttpReadOnly = true;
[[Overview]]\n[[People]]\n
!!!''Topics to be covered'' \n(this is only a first-draft and likely to change)\n\n#''Segmentation''\n##simple schemes (edge-based, threshold, region-grow, watershed,...)\n##EM, optical flow,...\n##graph partitioning based methods (network flow based and spectral methods)\n#''Representation and analysis'' \n##radial, polygonal, moments\n##skeletonization\n##graph-based data structures\n#''Registration''\n##simple methods for 2D\n##methods for rigid registration\n##Procrustes, ICPT\n##deformable contour models, level set methods\n#''Classification/Clustering''\n##intro to CAD\n##Representation and popular clustering schemes\n##Dimensionality reduction, near-neighbor searches, embedding\n##Support Vector Machines (?)\n#''Geometry/Topology''\n##Marching Cubes\n##Triangulation and surface decimation \n#''Presentations and Projects''\n\nProgramming component in a high level language.
!Matlab\n\n[[Quick overview |http://www.mathworks.com/access/helpdesk/help/techdoc/index.html?/access/helpdesk/help/techdoc/learn_matlab/bqr_2pl.html&http://www.cs.cmu.edu/afs/andrew/scs/cs/15-385/www/links.html]]\n[[A nice introduction to the Image Processing toolbox|http://www.mathworks.com/access/helpdesk/help/toolbox/images/images.shtml]]\n\n!C/C++/Java related\n\n[[CImg|http://cimg.sourceforge.net/]]\n[[VTK|http://www.vtk.org/]]\n[[ITK|http://www.itk.org/]]\n[[OpenCV|http://sourceforge.net/projects/opencvlibrary/]]\n[[MIPAV|http://mipav.cit.nih.gov/index.php]]\n[[CGAL|http://www.cgal.org/]]\n[[LEDA|http://www.algorithmic-solutions.com/leda/ledak/index.htm]]\n\n!Other Useful tools\n\n[[SPM|http://www.fil.ion.ucl.ac.uk/spm/]]\n[[FSL|http://www.fmrib.ox.ac.uk/fsl/]]\n[[AFNI|http://afni.nimh.nih.gov/afni/download/plonesoftwarecenter_view]]\n[[ImageJ|http://rsb.info.nih.gov/ij/]]\n[[Paraview|http://www.paraview.org/New/index.html]]\n[[GIMP|http://www.gimp.org/]]\n\n\n!Conferences and Journals of interest\n\n[[Transactions on Medical Imaging|http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=42]]\n[[Medical Image Analysis|http://www.elsevier.com/wps/find/journaldescription.cws_home/620983/description#description]]\n[[NeuroImage|http://www.sciencedirect.com/science/journal/10538119]]\n[[MICCAI|http://www.informatik.uni-trier.de/~ley/db/conf/miccai/index.html]]\n[[IPMI|http://www.informatik.uni-trier.de/~ley/db/conf/ipmi/index.html]]\n[[ISBI|http://www.biomedicalimaging.org/]]\n[[Medical Physics|http://scitation.aip.org/medphys/]]\n[[J. of MRI|http://www3.interscience.wiley.com/journal/10005199/home]]\n\n\n[[ICCV|http://www.informatik.uni-trier.de/~ley/db/conf/iccv/index.html]]\n[[CVPR|http://www.informatik.uni-trier.de/~ley/db/conf/cvpr/index.html]]\n[[ECCV|http://www.informatik.uni-trier.de/~ley/db/conf/eccv/index.html]]\n[[International J. of Computer Vision|http://www.springerlink.com/content/0920-5691]]\n[[Pattern Recognition|http://www.informatik.uni-trier.de/~ley/db/journals/pr/index.html]]\n[[PAMI|http://www.computer.org/portal/site/transactions/menuitem.802944db300bb678c4f34b978bcd45f3/index.jsp?&pName=tpami_home&]]\n[[NIPS|http://books.nips.cc/]]\n
Psychology Bldg. Room 115 @ 1:00PM-2:15PM, Tuesdays & Thursdays.\n\nFor a campus map, go [[here|http://www.map.wisc.edu]]\n\n
[[Class1]]\n[[Class2]]\n[[Class3]]\n[[Class4]]\n[[Class5]]\n[[Class6]]\n[[Class7]]\n[[Class8]]\n[[Class9]]\n[[Class10]]\n[[Class11]]\n[[Class12]]\n[[Class13]]\n[[Class14]]\n[[Class15]]\n[[Class16]]\n[[Class17]]\n[[Class18]]\n[[Class19]]\n[[Class20]]\n[[Class21]]\n[[Class22]]\n[[Class23]]\n[[Class24]]\n[[Class25]]\n[[Class26]]\n[[Class27]]\n[[Class28]]\n[[Class29]]\n[[Class30]]\n\nThe files section is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09]]
!!''CS 638 Special Topics, Computational Methods in Medical Image Analysis (Spring 2009)''\n\n!!!''Overview''\n\nThis course will introduce us to medical image analysis algorithms. We will cover topics such as registration, segmentation, classification and clustering with a focus on biomedical image data. By the end of this course, we will gain a good understanding of the current problems in biomedical image analysis, the techniques employed to address such problems as well as outstanding research issues.\n\n!!!''Prerequisites''\n\nYou should be comfortable with a high level programming language (C, C++, or Java) or Matlab/Octave. Basic familiarity with linear algebra and calculus will be assumed. Come talk to me if you are in doubt about any other aspects of the course. \n\n!!!''Location/Times''\n\nThe class will meet in Psychology Bldg. Room 115 @ 1:00PM-2:15PM, Tuesdays & Thursdays.\n\n!!!''Grading''\n\nWill be based on programming assignments and/or homeworks, reading assignments, presentations, and individual or team projects. No exams.\n\nHere is a [[preliminary list of topics]] we plan to cover. Also see [[reference textbooks]] and [[Readings]].\n\nA shorter version of this course announcement can be found [[here|http://www.biostat.wisc.edu/~vsingh/courseDescription638.pdf]]\n\n\n
# Welcome\n# Introduction to the course: problem types, data, topics to be covered, organization\n\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture1.pdf]]
#Connectivity basics\n#Properties and characteristics of regions \n#Some examples of uses of Euler Number\n#Brief introduction to and demo of \n##Convex Hulls\n##Medial axes\n##Voronoi diagrams\n##Triangulations and mesh construction\n#Defining distances (Metric property)\n#Image representation as \n##Matrices\n##Graphs\n#Example of boundary representation as chain code\n##Notion of invariance\n##Simplification and preprocessing \n\n\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture2.pdf]]
!Instructor\n\n''[[Vikas Singh| http://www.biostat.wisc.edu/~vsingh]]''\n\n7655 Medical Sciences Center\n1300 University Ave.,\nMadison, WI 53706.\n\nPhone: (608)262-8875.\nemail: username is vsingh, domain is biostat DOT wisc DOT edu\n\n''Office hours:'' Tuesdays/Thursdays, 4:00 - 5:00 PM. You are free to drop by at other times if I am in.\n\n!Teaching Assistant\n\n''Shengnan Wang''\n\n1307 CS & Statistics\n1210 W. Dayton St.,\nMadison, WI 53706.\n\nPhone: (608)xxx-xxxx\nemail: username is shengnan, domain is cs DOT wisc DOT edu\n\n''Office hours:'' Wednesdays, 4:00 - 5:00 PM.\n\n
#Pyramidal and hierarchical representations\n#Histograms and histogram operations\n#Morphological operations\n#Masking, filtering, estimating derivatives\n#Masks for edges\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture3.pdf]]
#Finding edges\n##Different types of edge operators\n#Threshold based\n##Otsu's method\n##Gaussian case\n#Region based and watershed\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture4.pdf]]
There are no required textbooks. However, you may find some of the topics in the following books useful.\n\n#Insight into Images: Principles & Practice for Segmentation, Registration and Image Analysis, Terry S. Yoo (will be ''On reserve at Wendt'' soon)\n#N Paragios, Y Chen, O Faugeras, Handbook of Mathematical Models in Computer Vision\n#Atam P. Dhawan, Medical Image Analysis (''On reserve at Wendt'')\n#Rangaraj M. Rangayyan, Biomedical Image Analysis (''Available online through UW''), go [[here|http://www.library.wisc.edu/]] \n#Image Processing, Analysis, and Machine Vision, M. Sonka, V. Hlavac, and R. Boyle.
#Model based segmentation (using Hough transform)\n##For arbitrary shapes\n##Examples\n#Intelligent scissors\n##Setting up graph and computing edge costs\n##Shortest path and Implementation issues\n#Brief intro to active contours/snakes\n\nShortest path slides are [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files0/demo_sp.pdf]]\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture5.pdf]]
#Intro to deformable segmentation\n#Review of variational calculus and Euler Lagrange eqn\n#Snakes energy term construction\n##external, internal and the influence of weights\n#gradient descent method\n#configuration space and greedy strategies\n#approximation with pairwise and triple potentials\n#motivate dynamic programming\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture6Part1.pdf]]
n/a
#Review of greedy method for snakes\n#Expansion of snakes energy as pairwise and triples\n#Dynamic Programming example\n#DP for solving snakes\n##pairwise \n##triples\n#Overview of level sets\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture6Part2.pdf]]
[[Intelligent Scissors for Image Composition|http://www.biostat.wisc.edu/~vsingh/teaching/files09/scissors_comp.pdf]], E. N. Mortensen and W. A. Barrett, SIGGRAPH 1995. \n\n[[Dynamic programming for detecting, tracking, and matching deformable contours|http://www.biostat.wisc.edu/~vsingh/teaching/files09/geiger.pdf]], D. Geiger, A. Gupta, L. A. Costa, and J. Vlontzos, IEEE Tran. on Pattern Analysis and Machine Intelligence 1995.\n\n[[Using dynamic programming for solving variational problems in Vision|http://www.biostat.wisc.edu/~vsingh/teaching/files09/amini.pdf]], A. Amini, T. E. Weymouth, R. C. Jain, IEEE Tran. on Pattern Analysis and Machine Intelligence 1990.\n\n[[Geodesic active contours|http://www.biostat.wisc.edu/~vsingh/teaching/files09/caselles95geodesic.pdf]], V. Caselles, R. Kimmel, and G. Sapiro, International Journal of Computer Vision 1997.\n\n[[Region tracking on level-sets methods|http://www.biostat.wisc.edu/~vsingh/teaching/files09/bertalmio.pdf]], M. Bertalmio, G. Sapiro, and G. Randall, IEEE Tran. Medical Imaging 1999.\n\n[[An efficient algorithm for image segmentation, Markov Random Fields and related problems|http://www.biostat.wisc.edu/~vsingh/teaching/files09/segment.pdf]], D. Hochbaum, Journal of ACM 2001.\n\n[[Object recognition from local scale-invariant features|http://www.biostat.wisc.edu/~vsingh/teaching/files09/sift.pdf]], D. G. Lowe, International Conference on Computer Vision 1999. \n\n[[Fast Approximate Energy Minimization via Graph Cuts|http://www.biostat.wisc.edu/~vsingh/teaching/files09/BVZ-pami01.pdf]], Y. Boykov, O. Veksler, and R. Zabih. IEEE Tran. on Pattern Analysis and Machine Intelligence 2001.\n\n[[Normalized Cuts and Image Segmentation|http://www.biostat.wisc.edu/~vsingh/teaching/files09/SM-ncut.pdf]], J. Shi and J. Malik, IEEE Tran. on Pattern Analysis and Machine Intelligence 2000.\n\n[[User-Steered Image Segmentation Paradigms: Live Wire and Live Lane|http://www.ingentaconnect.com/content/ap/ip/1998/00000060/00000004/art00475]], A. X. Falcao, J. K. Udupa, S. Samarasekera, S. Sharma, B. E. Hirsch, A.D.R. Lotufo, Graphical Models and Image Processing 1998.\n\n[[Predicting error in rigid-body point-based registration|http://www.biostat.wisc.edu/~vsingh/teaching/files09/fitzpatrick.pdf]], J. M Fitzpatrick, J. B West, C. R Maurer Jr., IEEE Tran. on Medical Imaging 1998.\n\n[[A new point matching algorithm for non-rigid registration|http://www.biostat.wisc.edu/~vsingh/teaching/files09/rangarajan.pdf]], H. Chui and A. Rangarajan, Computer Vision and Image Understanding 2003. \n\n[[Landmark Matching Via Large Deformation Diffeomorphisms|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Joshi_TIP_2000.pdf]], S. Joshi and M. Miller, IEEE Tran. on Image Processing 2000.\n\n[[Geodesic Active Regions and Level Set Methods for Supervised Texture Segmentation|http://www.biostat.wisc.edu/~vsingh/teaching/files09/levelset_ijcv02.pdf]], N. Paragios and R. Deriche, International Journal of Computer Vision 2002. \n\n[[Multimodality Image Registration by Maximization of Mutual Information|http://www.biostat.wisc.edu/~vsingh/teaching/files09/maes.pdf]], F. Maes, A. Collignon, D. Vandermeulen, G. Marchal, and P. Suetens, IEEE Tran. on Medical Imaging 1997.\n\n[[Mutual information based registration of medical images|http://www.biostat.wisc.edu/~vsingh/teaching/files09/mutual_info_survey.pdf]], J. P. W. Pluim, J. B. A. Maintz, and M. Viergever, IEEE Tran. on Medical Imaging 2003. \n\n[[Snakes: Active Contour Models|http://www.biostat.wisc.edu/~vsingh/teaching/files09/snakesTerzopoulos.pdf]], M. Kass, A. Witkin, and D. Terzopoulos, International Journal of Computer Vision 1988.\n\n[[Active appearance models|http://www.biostat.wisc.edu/~vsingh/teaching/files09/cootes.pdf]], T. F. Cootes, G. J. Edwards, and C. J. Taylor, IEEE Tran. on Pattern Analysis and Machine Intelligence 2001. \n\n[[Shape modeling with front propagation: a level set approach|http://www.biostat.wisc.edu/~vsingh/teaching/files09/malladi.pdf]], R. Malladi, J. A. Sethian, B. C. Vemuri, IEEE Tran. on Pattern Analysis and Machine Intelligence 1995. \n\n[[Image processing via level set curvature flow|http://www.biostat.wisc.edu/~vsingh/teaching/files09/sethianpnas.pdf]], R. Malladi and J. A. Sethian, Proc. of National Acad. of Sciences, 1995.\n\n\n[[Deformable models in medical image analysis: a survey|http://www.biostat.wisc.edu/~vsingh/teaching/files09/survey_deformable.pdf]], T. McInerney and D. Terzopoulos, Medical Image Analysis 1996. \n\n[[Marching Cubes : A High Resolution 3D Surface Construction Algorithm|http://www.biostat.wisc.edu/~vsingh/teaching/files09/lorensen.pdf]], W. Lorensen and H. Cline, ACM SIGGRAPH 1987.\n\n[[Medical Image Registration|http://www.biostat.wisc.edu/~vsingh/teaching/files09/hill.pdf]], D. Hill, P. Batchelor, M. Holden, and D. J. Hawkes, Phys. Med. Biol. 2001.\n\n[[Image Registration|http://www.biostat.wisc.edu/~vsingh/teaching/files09/registration_chapter.pdf]], J. M. Fitzpatrick, D. Hill, C. R. Maurer Jr. (book chapter).\n\n[[Medical image analysis: progress over two decades and the challenges ahead|http://www.biostat.wisc.edu/~vsingh/teaching/files09/duncan.pdf]], J. Duncan and N. Ayache, IEEE Tran. on Pattern Analysis and Machine Intelligence 2000. \n\n[[HAMMER: Hierarchical Attribute Matching Mechanism for Elastic Registration|http://www.biostat.wisc.edu/~vsingh/teaching/files09/hammer.pdf]], D. Shen and C. Davatzikos, IEEE Tran. on Medical Imaging 2002. \n\n[[Snakes, Shapes, and Gradient Vector Flow|http://www.biostat.wisc.edu/~vsingh/teaching/files09/xuprince.pdf]], C. Xu and J. L. Prince, IEEE Tran. on Image Processing 1998.\n\n[[Edge Detection and Ridge Detection with Automatic Scale Selection|http://www.biostat.wisc.edu/~vsingh/teaching/files09/lindeberg.pdf]], T. Lindeberg, International Journal of Computer Vision 1998.\n\n[[A Survey of Medical Image Registration|http://www.biostat.wisc.edu/~vsingh/teaching/files09/maintz.pdf]], J. B. A. Maintz and M. A. Viergever, Medical Image Analysis 1998.\n\n[[Shape modeling with front propagation: a level set approach|http://www.biostat.wisc.edu/~vsingh/teaching/files09/malladi.pdf]], R. Malladi, J. Sethian, B. Vemuri, IEEE Tran. on Pattern Analysis and Machine Intelligence 1995. \n\n[[Optimum Image Thresholding via Class Uncertainty and Region Homogeneity|http://www.biostat.wisc.edu/~vsingh/teaching/files09/saha.pdf]], P. K. Saha and J. K. Udupa, IEEE Tran. on Pattern Analysis and Machine Intelligence 2001. \n\n[[Efficient multilevel image thresholding|http://www.biostat.wisc.edu/~vsingh/teaching/files09/multilevelImageThresholding.pdf]], M. Eichmann, M. Lussi, 2005. \n\n[[Minimum error thresholding|http://portal.acm.org/citation.cfm?id=20363]], J. Kittler and J. Illingworth, Pattern Recognition 1986.\n\n[[A Gentle Tutorial of the EM Algorithm and its Application to Parameter Estimation for Gaussian Mixture and Hidden Markov Models|http://www.biostat.wisc.edu/~vsingh/teaching/files09/blimes.pdf]], J. A. Blimes, ICSI Technical Report 1997.\n\n[[Mean shift, mode seeking and clustering|http://www.biostat.wisc.edu/~vsingh/teaching/files09/cheng.pdf]], Y. Chen, IEEE Tran. on Pattern Analysis and Machine Intelligence 1995.\n\n[[Mean Shift Analysis and Applications|http://www.biostat.wisc.edu/~vsingh/teaching/files09/comaniciu.pdf]], D. Comaniciu and P. Meer, International Conference on Computer Vision 1999. \n\n[[A method for the registration of 3-D shapes|http://www.biostat.wisc.edu/~vsingh/teaching/files09/besl.pdf]], P. Besl and N. McKay, IEEE Tran. on Pattern Analysis and Machine Intelligence 1992.\n\n[[Multi-resolution elastic matching|http://www.biostat.wisc.edu/~vsingh/teaching/files09/bajcsy.pdf]], R. Bajcsy and S. Kovacic, Computer Vision, Graphics and Image Processing 1989. \n\n[[Understanding the Demon's Algorithm: 3D Non-rigid Registration by Gradient Descent|http://www.biostat.wisc.edu/~vsingh/teaching/files09/pennec.pdf]], X. Pennec, P. Cachier, and N. Ayache, Medical Image Computing and Computer-Assisted Intervention 1999. \n\n
#Level Sets\n##Intro and differences with ODE\n##Convection\n##Signed Distance functions\n##Speed and curvature\n##Stationary version: Eikonal\n##Fast Marching and discretization\n#Intro to Max-flow\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture6Part3.pdf]]\n\nAdditional recommended readings: \n#[[Shape modeling with front propagation: a level set approach|http://www.biostat.wisc.edu/~vsingh/teaching/files09/malladi.pdf]]\n#[[Image processing via level set curvature flow|http://www.biostat.wisc.edu/~vsingh/teaching/files09/sethianpnas.pdf]]\n\n
1. overview of Markov Random Field\n2. overview of seeded segmentation MRF based segmentation energy function\n3. Graph construction for \salpha-\sbeta swap\n4. Graph construction for \salpha-expansion\n5. Examples\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture6Part4.pdf]]\n\nAdditional recommended readings: \n#[[Fast Approximate Energy Minimization via Graph Cuts|http://www.biostat.wisc.edu/~vsingh/teaching/files09/BVZ-pami01.pdf]]
#Review of Graph Cuts\n##\salpha-expansion graph construction\n#Examples\n#Relationship to variational methods\n#Random Walk based segmentation\n##Overview\n##Formulation/derivation using Laplacian\n##Analogy to other problems\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture6.pdf]]\n\nAdditional recommended readings: \n#[[Fast Approximate Energy Minimization via Graph Cuts|http://www.biostat.wisc.edu/~vsingh/teaching/files09/BVZ-pami01.pdf]]\n#[[Random Walks for Image Segmentation|http://www.biostat.wisc.edu/~vsingh/teaching/files09/grady2006random.pdf]]
#Review of Random Walks\n#Feature Extraction \n##Harris\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture12Part1.pdf]]
#Review of Harris corner detectors\n##Limitations\n#SIFT features\n##construction procedure\n##examples and demo\n#Image registration\n##Intro and broad issues\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture12.pdf]]\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture13Part1.pdf]]
No class.
#Review of ICP\n#Entropy and Randomness\n#Relationship of entropy to KL-divergence and upper bound\n#Conditional entropy\n#Mutual Information in terms of entropy\n#Joint image histogram\n#Examples\n#Some optimization schemes for MI\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture13Part3.pdf]]
#Similarity measures for registration\n#Rigid registration\n##Rotation, translation, scaling,..\n#Finding rigid transformation\n#Principal axis based registration using eigen-values/eigen-vectors\n#Intro to Iterative Closest point (ICPT)\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture13Part2.pdf]]
#Intro to Deformation registration\n#Brief review of Splines for interpolating data\n#Thin Plate Splines based registration\n##Cost function, regularizer, warping\n#Chui/Rangarajan: EM style TPS registration without correspondences
#Review of Thin Plate Splines Registration\n##Examples and demos\n#Demons based Image registration\n##Key ideas: Maxwell's demons\n##Optical Flow: ideas and demos\n##How to use optical flow for Demons registration\n##Examples\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture13.pdf]]
Guest Lecture by [[Prof. Moo K. Chung|http://www.stat.wisc.edu/~mchung/]]\n\nTitle: Eigenfunctions of Laplace Beltrami operator on cortical manifolds\n\nIn quantifying cortical and subcortical anatomy of the human brain, various differential geometric methods have been proposed. Many such successful methods are inherently implicit and without explicit parametric forms. Although there are few parametric approaches such as spherical harmonic descriptors, the application has been limited to simple subcortical structures. The reason for the lack of more explicit parametric approaches is that it is difficult to construct an orthonormal basis for an arbitrary cortical manifold. We propose to use the eigenfunctions of the Laplace-Beltrami operator, which are computed numerically using the cotan fomula. The eigenfunctions are then used in setting up a regression in the cortical manifold. The eigenfunction approach offers far more flexibility in setting up a statistical model than implicit approaches. However, the eigenfunction approach suffers some inherent technical issues that are yet to be resolved. We will discuss the limitation of the eigenfunction method.\n\nPDF slides are [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/moopresmarch_31_2009.pdf]] ''18MB!''
#Intro to Machine learning in biomedical image processing\n#Supervised and Unsupervised: clustering and classification\n#Utility of model compexity, bias, priors\n#k-means\n##discussion of solvability\n#fuzzy variations of clustering\n\n\nPDF slides are [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture21Part1.pdf]]
#Affinity Propagation\n##Overview\n##Types of messages\n##Limitations and advantages\n#Overview of Spectral Clustering\n##Model construction\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture21Part2.pdf]]
#Spectral Clustering\n##Review\n##Objective function and eigen value problem\n##Variations (Ng et al.)\n#Bayesian classification\n##Calculating conditional, joint etc\n##Naive Bayes\n###examples\n#k-nearest neighbor\n##Introduction\n##choosing k using cross-validation\n##efficient nearest neighbor searches\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture21Part3.pdf]]
#Review of k-nearest neighbors\n#Motivating linear separability\n#Perceptrons\n#Simple dot-product based classification\n#maximum margin principle\n#Kernel trick\n#Geometric derivation of SVM primal model\n#Lagrangian and dual\n#Examples to diagnosis in brain images and breast cancer\n\nPDF is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Lecture21Part4.pdf]]
#Remainder of Clustering/Classification lectures\n#Presentations\n\n''Presentation by Jake Rosin'' (also see [[Projects]])\nTitle: ''Volume visualization in biomedical imaging''\nAbstract: Many modern medical imaging techniques produce volumetric data. Although volumetric data such as that produced by MRI or CT techniques include strictly more information than traditional 2D medical radiography, there is no clearly correct method for rendering that data that does not abstract away certain details. Rendering techniques therefore necessarily sacrifice some aspect of the information in order to emphasize another. In some cases, realism and spatial accuracy are acceptable loses if the resulting image clearly conveys some important aspect of the data. This presentation examines a few techniques for volume visualization that produce stylized, or in some other way nonrealistic, output, and the reasons why such output might be preferred. \n\nPDF of presentation is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/JakeRosin.pdf]]\n\nRelated readings:\n#Armin Kanitsar, Dominik Fleischmann, Rainer Wegenkittl, Petr Felkel, Meister Eduard Groller, CPR - Curved Planar Reformation.\n#Armin Kanitsar, Rainer Wegenkittl, Dominik Fleischmann, Meister Eduard Groller, Advanced Curved Planar Reformation: Flattening of Vascular Structures. \n#Michael Burns, Janek Klawe, Szymon Rusinkiewicz, Adam Finkelstein, Doug DeCarlo, Line Drawings from Volume Data.\n#Ivan Viola, Armin Kanitsar, Meister Eduard Gröller, Importance-Driven Volume Rendering.\n#Nikolai A. Svakhine, David S. Ebert, William M. Andrews, Illustration-Inspired Depth Enhanced Volumetric Medical Visualization.\n\n
''Tentative Presentation Schedule''\n\n#''Rosin'' (21 April 2009)\n#''Song'' (23 April 2009)\n#''Diamond'' (23 April 2009)\n#''Wang'' (28 April 2009)\n#''Doot'' (28 April 2009)\n#''Rastegar'' (30 April 2009)\n#''Maynord'' (30 April 2009)\n#''Chiang'' (5 May 2009)\n#''Mohammad'' (5 May 2009)\n#''Printz'' (7 May 2009)\n\n''Project titles''\n\n''Rosin'': Volume visualization in biomedical imaging\nMany modern medical imaging techniques produce volumetric data. Although volumetric data such as that produced by MRI or CT techniques include strictly more information than traditional 2D medical radiography, there is no clearly correct method for rendering that data that does not abstract away certain details. Rendering techniques therefore necessarily sacrifice some aspect of the information in order to emphasize another. In some cases, realism and spatial accuracy are acceptable loses if the resulting image clearly conveys some important aspect of the data. This presentation examines a few techniques for volume visualization that produce stylized, or in some other way nonrealistic, output, and the reasons why such output might be preferred. \n\n''Song'': Implementation of Demons and optical flow based image registration\ndescription here\n\n''Diamond'': Computer Aided Detection of Bone Metastases in CT images\nBone metastasis is the spread of cancer cells to bone from another site in the body. When the cancer cells attach themselves to bone and begin to grow they form lesions which damage the surrounding bone structure and significantly limit a patient’s quality of life as well as their life expectancy. Bone metastasis appear as round holes or whiter regions inside of bones in the axial skeleton during routine CT scans, but can be missed by radiologists focusing on other aspects of the scans. My Project is to devise an automatic detection algorithm which can run in the background on routine CTs to alert radiologists to the possibility of metastatic disease in patients.\n\n''Wang'': Multiclass Learning by Boosting Boostrap FLD Subspaces\nAn ensemble feature extraction algorithm will be implemented based on Adaboost.M2 for multiclass classification problem. The method will first sample a large number of bootstrap training subsets from the original training set and implements FLD in each subset to get a large number of bootstrap FLD projections. Then at each step of\nAdaboost.M2, the projection with the minimum weighted K-nearest Neighbor (KNN) classification error will be selected from a pool of linear projections to combine the final strong classifier.\n\n''Doot'': Segmentation of Biological Multiphoton data\nThe techniques introduced in this talk allow for accurate multiscale image reconstruction of multi-photon microscopy data. Multi-photon microscopy (MPM) is a tool for the imaging of living organisms and tissue. The data acquired using this technique contain information about the position, excited state lifetime, and spectra of the observed photons. The small number of photons collected, however, limits the quality of the reconstructions. The multiscale framework in this talk results in an accurate representation of both the intensity and excited state lifetime information. Using a multiscale reconstruction approach based on a penalized likelihood function, the underlying image is more accurately represented as compared to a naive aggregate binning approach. \n\n''Rastegar''\ntitle, description here\n\n''Maynord''\ntitle, description here\n\n''Chiang''\ntitle, description here\n\n''Mohammad''\ntitle, description here\n\n''Printz''\ntitle, description here
''Presentation by Jiasi Song'' (also see [[Projects]])\nTitle: ''Demons and optical flow based image registration''\nAbstract: TBA\nPDF of presentation is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/JiasiSong.pdf]]\n\nRelated readings:\n#Andres Bruhn, Joachim Weickert, Christoph Schnorr, Lucas/Kanade meets horn/schunck: Combining local and global optic flow methods.\n#J. P. Thirion, Image matching as a diffusion process: an analogy with Maxwell’s demons.\n\n''Presentation by Greg Diamond'' (also see [[Projects]])\nTitle: ''Computer Aided Detection of Bone Metastases''\nAbstract: Bone metastasis is the spread of cancer cells to bone from another site in the body. When the cancer cells attach themselves to bone and begin to grow they form lesions which damage the surrounding bone structure and significantly limit a patient’s quality of life as well as their life expectancy. Bone metastasis appear as round holes or whiter regions inside of bones in the axial skeleton during routine CT scans, but can be missed by radiologists focusing on other aspects of the scans. My Project is to devise an automatic detection algorithm which can run in the background on routine CTs to alert radiologists to the possibility of metastatic disease in patients.\nPDF of presentation is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/GregDiamond.pdf]]\n\nRelated readings:\n#Jianhua Yao, Stacy D. O’Connor, Ronald M. Summers, Computer Aided Detection of Lytic Bone Metastases in the Spine using Routine CT Images.
''Presentation by Tuo Wang'' (also see [[Projects]])\nTitle: ''Multiclass Learning by Boosting Boostrap FLD Subspaces''\nAbstract: An ensemble feature extraction algorithm will be implemented based on Adaboost.M2 for multiclass classification problem. The method will first sample a large number of bootstrap training subsets from the original training set and implements FLD in each subset to get a large number of bootstrap FLD projections. Then at each step of Adaboost.M2, the projection with the minimum weighted K-nearest Neighbor (KNN) classification error will be selected from a pool of linear projections to combine the final strong classifier.\n\nPDF of presentation is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/TuoWang.pdf]]\n\nRelated readings:\n#D. Masip, J. Vitria, Boosted discriminant projections for nearest\nneighbor classification.\n#Y. Freund, R. E. Schapire, A decision-theoretic generalization of\nonline learning and an application to boosting.\n#J. Lu, K. N. Plataniotis, A. N. Venetsanopoulos, Boosting linear\ndiscriminant analysis for face recognition.\n#D. Masip, L. Kuncheva, J. Vitria, An ensemble-based method for linear feature extraction for two-class problems.\n\n\n''Presentation by Jared Doot'' (also see [[Projects]])\nTitle: ''Segmentation of Biological Multiphoton data''\nAbstract: The techniques introduced in this talk allow for accurate multiscale image reconstruction of multi-photon microscopy data. Multi-photon microscopy (MPM) is a tool for the imaging of living organisms and tissue. The data acquired using this technique contain information about the position, excited state lifetime, and spectra of the observed photons. The small number of photons collected, however, limits the quality of the reconstructions. The multiscale framework in this talk results in an accurate representation of both the intensity and excited state lifetime information. Using a multiscale reconstruction approach based on a penalized likelihood function, the underlying image is more accurately represented as compared to a naive aggregate binning approach. \n\nPDF of presentation is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/JaredDoot.pdf]]\n\nRelated readings:\n#E. D. Kolaczyk and R. D. Nowak, Multiscale Likelihood Analysis and Complexity Penalized Estimation. \n#R. Willett and R. Nowak, Multiscale Poisson Intensity and Density Estimation.
Guest Lecture by [[Prof. Andy Alexander|http://www.waisman.wisc.edu/faculty/alexander.html]]\n\nTitle: Diffusion MRI: Tensors and Beyond\n\nRecommended Readings: [[Paper 1|http://www.biostat.wisc.edu/~vsingh/teaching/files09/Alexander.DTI.Neurother.7.07.pdf]] and [[Paper 2|http://www.biostat.wisc.edu/~vsingh/teaching/files09/jones_dtireview_cortex2008.pdf]]\n\nPDF slides are [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/DTIMethods022109.pdf]]
''Presentation by Michael Maynord'' (also see [[Projects]])\nTitle: ''Random Walks''\nAbstract: Random walks are a powerful general purpose image segmentation method. This presentation will explain a bit about what they, how they work, and cover a clinical application. \n\nPDF of presentation is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/MichaelMaynord.pdf]]\n\nRelated readings:\n#L. Grady, Random Walks for Image Segmentation.\n\n''Presentation by Ted Chiang'' (also see [[Projects]])\nTitle: ''Level sets based segmentation''\nAbstract: \n\nPDF of presentation is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/TedChiang.pdf]]\n\nRelated readings:\n#R. Malladi, J. Sethian, B. Vemuri, Shape modeling with front propagation: A level set approach
''Presentation by Jared Doot'' (also see [[Projects]])\nTitle: ''Segmentation of Biological Multiphoton data''\nAbstract: The techniques introduced in this talk allow for accurate multiscale image reconstruction of multi-photon microscopy data. Multi-photon microscopy (MPM) is a tool for the imaging of living organisms and tissue. The data acquired using this technique contain information about the position, excited state lifetime, and spectra of the observed photons. The small number of photons collected, however, limits the quality of the reconstructions. The multiscale framework in this talk results in an accurate representation of both the intensity and excited state lifetime information. Using a multiscale reconstruction approach based on a penalized likelihood function, the underlying image is more accurately represented as compared to a naive aggregate binning approach. \n\nPDF of presentation is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/JaredDoot.pdf]]\n\nRelated readings:\n#E. D. Kolaczyk and R. D. Nowak, Multiscale Likelihood Analysis and Complexity Penalized Estimation. \n#R. Willett and R. Nowak, Multiscale Poisson Intensity and Density Estimation.\n\n''Presentation by Farzad Rastegar'' (also see [[Projects]])\nTitle: ''Background Modeling Using Probabilistic Approaches''\nAbstract: Real-time visual surveillance systems are used for numerous applications such as detecting and tracking people and behavior analysis. The most significant issue in these systems is how to segment moving objects because the better the segmentation of moving objects is, the more fruitful information we can obtain from the images. A common approach for real-time segmentation of moving objects in image sequences involves background subtraction or thresholding the error between an estimate of the image without moving objects and the current image. Probabilistic approaches are among promising techniques for background modeling. In this talk, we introduce two probabilistic methods and show how these techniques can be utilized for this purpose. First, we take a look at the uniform distribution and explain how it can be used for segmentation of moving objects. Then, we elaborate a more complicated model which utilizes the mixture of Gaussians to come up with an adaptive technique for background estimation.\n\nPDF of presentation is [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/FarzadRastegar.pdf]]\n\nRelated readings:\n#C. Stauffer, W. E. L. Grimson, Learning patterns of activity using real-time tracking.\n
#First class meets on 01/20/2009, Tuesday\n#Lecture 1 slides posted (1/20/2009).\n#Lecture 2 slides posted (1/22/2009).\n#Lecture 3 slides posted (1/27/2009).\n#Lecture 4 slides posted (1/29/2009).\n#Lecture 5 slides posted (2/03/2009).\n#Lecture 6 slides posted (2/05/2009).\n#Lecture 7 slides posted (2/10/2009).\n#Lecture 9 slides posted (2/17/2009).\n#Lecture 10 slides posted (2/19/2009).\n#''HW 1 is [[posted|http://www.biostat.wisc.edu/~vsingh/teaching/files09/hw1.pdf]]'' (due on ''March 5, 2009, 11:59PM'').\n#Lecture 11 slides posted (2/24/2009).\n#Lecture 12 slides posted (2/26/2009).\n#Lecture 13 slides posted (3/03/2009).\n#Lecture 14 slides posted (3/05/2009).\n#''No class on 03/10/2009''.\n#Lecture 16 slides posted (3/12/2009).\n#''1-page Project proposal is due on'' ==3/27/2009== ''3/31/2009'' \n#Lecture 17-18 slides posted (3/26/2009).\n#''HW 2 is [[posted|http://www.biostat.wisc.edu/~vsingh/teaching/files09/hw2.pdf]]'' (due on ''April 14, 2009, 11:59PM'').\n#Lecture 19 slides (Moo K. Chung lecture) posted (3/31/2009).\n#Lecture 21 slides posted (4/07/2009).\n#Lecture 20 slides (Andy L. Alexander lecture) posted (4/09/2009).\n#Lecture 22 slides posted (4/09/2009).\n#Lecture 23 slides posted (4/14/2009).\n#Lecture 24 slides posted (4/19/2009).\n#''HW 3 is [[posted|http://www.biostat.wisc.edu/~vsingh/teaching/files09/hw3.pdf]]'' (due on ''May 5, 2009, 11:59PM'').\n#''Project reports/code due on May 8, 2009, 11:59PM'' (submission instructions same as homeworks using handin, the directory is projects rather than hw3). I have a template available [[here|http://www.biostat.wisc.edu/~vsingh/teaching/files09/tr-template.zip]]. Look [[here|http://dimacs.rutgers.edu/TechnicalReports/submit.html]] to download some other example templates which are fine too. The code should be well documented and must come with a README.