Medical Image Analysis Dr. Mohammad Dawood Department of Computer Science University of Münster Germany.

Slides:



Advertisements
Similar presentations
Image Registration  Mapping of Evolution. Registration Goals Assume the correspondences are known Find such f() and g() such that the images are best.
Advertisements

3D Geometry for Computer Graphics
The fundamental matrix F
Medical Image Registration Kumar Rajamani. Registration Spatial transform that maps points from one image to corresponding points in another image.
Medical Imaging Mohammad Dawood Department of Computer Science University of Münster Germany.
PCA + SVD.
1er. Escuela Red ProTIC - Tandil, de Abril, 2006 Principal component analysis (PCA) is a technique that is useful for the compression and classification.
Computer Vision Optical Flow
Reverse Engineering Niloy J. Mitra.
Principal Component Analysis CMPUT 466/551 Nilanjan Ray.
Medical Imaging Mohammad Dawood Department of Computer Science University of Münster Germany.
Medical Imaging Mohammad Dawood Department of Computer Science University of Münster Germany.
Principal Component Analysis
Mosaics con’t CSE 455, Winter 2010 February 10, 2010.
CENG 789 – Digital Geometry Processing 06- Rigid-Body Alignment Asst. Prof. Yusuf Sahillioğlu Computer Eng. Dept,, Turkey.
Structure from motion. Multiple-view geometry questions Scene geometry (structure): Given 2D point matches in two or more images, where are the corresponding.
CSc83029 – 3-D Computer Vision/ Ioannis Stamos 3-D Computational Vision CSc Optical Flow & Motion The Factorization Method.
Mutual Information for Image Registration and Feature Selection
The Terms that You Have to Know! Basis, Linear independent, Orthogonal Column space, Row space, Rank Linear combination Linear transformation Inner product.
Fitting a Model to Data Reading: 15.1,
3D Geometry for Computer Graphics
COMP 290 Computer Vision - Spring Motion II - Estimation of Motion field / 3-D construction from motion Yongjik Kim.
CS4670: Computer Vision Kavita Bala Lecture 7: Harris Corner Detection.
Lecture 10: Robust fitting CS4670: Computer Vision Noah Snavely.
כמה מהתעשייה? מבנה הקורס השתנה Computer vision.
Multi-view geometry. Multi-view geometry problems Structure: Given projections of the same 3D point in two or more images, compute the 3D coordinates.
Image Stitching Ali Farhadi CSE 455
Point set alignment Closed-form solution of absolute orientation using unit quaternions Berthold K. P. Horn Department of Electrical Engineering, University.
CSE 185 Introduction to Computer Vision
CSC 589 Lecture 22 Image Alignment and least square methods Bei Xiao American University April 13.
Computer Vision - Fitting and Alignment
Empirical Modeling Dongsup Kim Department of Biosystems, KAIST Fall, 2004.
Summarized by Soo-Jin Kim
CSE554AlignmentSlide 1 CSE 554 Lecture 8: Alignment Fall 2014.
CSE554Laplacian DeformationSlide 1 CSE 554 Lecture 8: Laplacian Deformation Fall 2012.
Multimodal Interaction Dr. Mike Spann
CSE554AlignmentSlide 1 CSE 554 Lecture 5: Alignment Fall 2011.
Course 12 Calibration. 1.Introduction In theoretic discussions, we have assumed: Camera is located at the origin of coordinate system of scene.
M Dawood1,2, C Brune2, F Büther1, KP Schäfers1
Medical Imaging Dr. Mohammad Dawood Department of Computer Science University of Münster Germany.
Medical Image Analysis Image Registration Figures come from the textbook: Medical Image Analysis, by Atam P. Dhawan, IEEE Press, 2003.
Computer Vision : CISC 4/689 Going Back a little Cameras.ppt.
CSE554AlignmentSlide 1 CSE 554 Lecture 8: Alignment Fall 2013.
Computer Vision - Fitting and Alignment (Slides borrowed from various presentations)
Using simplified meshes for crude registration of two partially overlapping range images Mercedes R.G.Márquez Wu Shin-Ting State University of Matogrosso.
EIGENSYSTEMS, SVD, PCA Big Data Seminar, Dedi Gadot, December 14 th, 2014.
Point Distribution Models Active Appearance Models Compilation based on: Dhruv Batra ECE CMU Tim Cootes Machester.
776 Computer Vision Jan-Michael Frahm Spring 2012.
Medical Image Analysis Dr. Mohammad Dawood Department of Computer Science University of Münster Germany.
CENG 789 – Digital Geometry Processing 07- Rigid-Body Alignment Asst. Prof. Yusuf Sahillioğlu Computer Eng. Dept,, Turkey.
MOTION Model. Road Map Motion Model Non Parametric Motion Field : Algorithms 1.Optical flow field estimation. 2.Block based motion estimation. 3.Pel –recursive.
EE 7730 Parametric Motion Estimation. Bahadir K. Gunturk2 Parametric (Global) Motion Affine Flow.
Lecture 10: Image alignment CS4670/5760: Computer Vision Noah Snavely
776 Computer Vision Jan-Michael Frahm Spring 2012.
CSE 554 Lecture 8: Alignment
CENG 789 – Digital Geometry Processing 08- Rigid-Body Alignment
University of Ioannina
Fitting a transformation: feature-based alignment
Lecture 7: Image alignment
A special case of calibration
CSE 554 Lecture 9: Laplacian Deformation
Image Stitching Slides from Rick Szeliski, Steve Seitz, Derek Hoiem, Ira Kemelmacher, Ali Farhadi.
Techniques for studying correlation and covariance structure
Recitation: SVD and dimensionality reduction
Image Registration 박성진.
Lecture 8: Image alignment
Day 05 Horn's Method 5/24/2019.
Calibration and homographies
Image Stitching Linda Shapiro ECE/CSE 576.
Image Stitching Linda Shapiro ECE P 596.
Presentation transcript:

Medical Image Analysis Dr. Mohammad Dawood Department of Computer Science University of Münster Germany

2 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Recap

3 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Sound wavesPiezoelectric crystalsWave front formation

4 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Inverse Radon transformFiltered back projection Filtered back projection

5 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Fourier slice theoremKaczmarz Method (=ART)

6 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Image Registration

7 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration T : Transformation In this lecture Floating image: The image to be registered Target image: The stationary image

8 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Linear Transformations - Translation - Rotation - Scaling - Affine

9 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration 3D Translation

10 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration 3D Rotation

11 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration 3D Scaling

12 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Rigid registration Angles are preserved Parallel lines remain parallel

13 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Affine registration

14 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Floating Target Registration Points cloud registration

15 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration PCA = Principle Components Analysis Compute a shape-aware coordinate system for each model Origin: Centroid of all points Axes: Directions in which the model varies most or least Transform the source to align its origin/axes with the target

16 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Principal Component Analysis Set of points p 1,…,p n with centroid location c Let P be a matrix whose i-th column is vector p i – c Build the covariance matrix: c F c T

17 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Principal Component Analysis Eigenvectors of M represent principal directions of shape variation Eigenvalues indicate amount of variation along each eigenvector Eigenvector with largest (smallest) eigenvalue is the direction where the model shape varies the most (least) Eigenvector with the largest eigenvalue Eigenvector with the smallest eigenvalue

18 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Noise Axes are affected PCA result Registration Principal Component Analysis Noise Symmetries Rotation by a small anglePCA result

19 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration SVD = Single Value Decomposition Assuming that for each source point the corresponding target point is known

20 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration SVD 1. De-mean 2. Compute SVD 3. Calculate the transform

21 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Translate Rotate

22 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration SVD Stable in noise Symmetries

23 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration ICP = Iterative closest point 1. Transform the floating image by PCA-based alignment 2. For each floating point, select the closest target point 3. Align floating and target images by SVD 4. Repeat steps 2 and 3 until error below threshold.

24 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood After PCA After 10 iterAfter 1 iter

25 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood After 10 iter After 1 iter After PCA

26 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Feature Points RANSAC = Random Sample Consensus Algorithm 1.Randomly choose s samples 2.Fit a model to those samples 3.Count the number of inliers that approximately fit the model 4.Repeat N times 5.Choose the model that has the largest set of inliers

27 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Distance Measures - Sum of Squared Differences (SSD) - Root Mean Square Difference (RMSD) - Normalized Cross Correlation (NXCorr) - Mutual Information (MI)

28 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Sum of Squared Differences SSD(f,t)SSD(20f,t)

29 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Root Mean Squared Differences RMS(f,t)RMS(20f,t)

30 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Normalized Cross Correlation NXCorr(f,t)NXCorr(20f,t)

31 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Registration Mutual Information MI(f,t)MI(20f,t)

32 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Entropy for Image Registration Define a joint probability distribution: – Generate a 2-D histogram where each axis is the number of possible greyscale values in each image – each histogram cell is incremented each time a pair (I 1 (x,y), I 2 (x,y)) occurs in the pair of images If the images are perfectly aligned then the histogram is highly focused. As the images mis-align the dispersion grows recall Entropy is a measure of histogram dispersion

33 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Optical Flow

34 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Optical flow Brightness consistency constraint With Taylor expansion V : Flow(Motion)

35 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood

36 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Optical flow Lucas Kanade Algorithm: Assume locally constant flow =>

37 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Optical flow Horn Schunck Algorithm: Assume globally smooth flow

38 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Optical flow Bruhn’s Non-linear Algorithm

39 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood PET no AC – gated PET no AC – gated, MC corrected

40 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Cardiac motion Ungated data Large blur, low noise (Problem in plaque imaging) One phase Small blur, high noise

41 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood SystoleDiastole Continuity Equation:

42 Medical Image Analysis, SS-2015 Dr. Mohammad Dawood Visual result