Upload
topaz
View
22
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Medical Imaging. Mohammad Dawood Department of Computer Science University of Münster Germany. Recap. Sound wavesPiezoelectric crystalsWave front formation. Inverse Radon transform Filtered back projection. Filtered back projection. - PowerPoint PPT Presentation
Citation preview
Medical Imaging
Mohammad Dawood
Department of Computer Science
University of MünsterGermany
2
Medical Imaging, SS-2011
Mohammad Dawood
Recap
3
Medical Imaging, SS-2011
Mohammad Dawood
Sound waves Piezoelectric crystals Wave front formation
4
Medical Imaging, SS-2011
Mohammad Dawood
Inverse Radon transform Filtered back projection
Filtered back projection
5
Medical Imaging, SS-2011
Mohammad Dawood
Fourier slice theorem Kaczmarz Method (=ART)
6
Medical Imaging, SS-2011
Mohammad Dawood
Image Registration
7
Medical Imaging, SS-2011
Mohammad Dawood
Registration
T : Transformation
In this lecture
Floating image : The image to be registered
Target image : The stationary image
8
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Linear Transformations
- Translation
- Rotation
- Scaling
- Affine
9
Medical Imaging, SS-2011
Mohammad Dawood
Registration
3D Translation
10
Medical Imaging, SS-2011
Mohammad Dawood
Registration
3D Rotation
11
Medical Imaging, SS-2011
Mohammad Dawood
Registration
3D Scaling
12
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Rigid registration
Angles are preserved Parallel lines remain parallel
13
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Affine registration
14
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Feature Points
15
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Feature Points
1. De-mean
2. Compute SVD
3. Calculate the transform
16
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Feature Points
Iterative Closest Points Algorithm (ICP)
1. Associate points by the nearest neighbor criteria.2. Estimate transformation parameters using a mean square cost function.3. Apply registration and update parameters.
17
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Feature Points
18
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Feature Points
Random Sample Consensus Algorithm (RNSAC)
1. Transformation is calculated from hypothetical inliers2. All other data are then tested against the fitted model and, if a point fits well to the model, also considered as a hypothetical inlier3. The estimated model is reasonably good if sufficiently many points have been classified as hypothetical inliers.4. The model is re-estimated from all assumed inliers5. Finally, the model is evaluated by estimating the error of the inliers relative to the model
19
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Phase Correlation
20
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Distance Measures
- Sum of Squared Differences (SSD)
- Root Mean Square Difference (RMSD)
- Normalized Cross Correlation (NXCorr)
- Mutual Information (MI)
21
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Sum of Squared Differences
SSD(f,t) SSD(20f,t)
22
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Root Mean Squared Differences
RMS(f,t) RMS(20f,t)
23
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Normalized Cross Correlation
NXCorr(f,t) NXCorr(20f,t)
24
Medical Imaging, SS-2011
Mohammad Dawood
Registration
Mutual Information
MI(f,t) MI(20f,t)
25
Medical Imaging, SS-2011
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 (I1(x,y), I2(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
26
Medical Imaging, SS-2011
Mohammad Dawood
Optical Flow
27
Medical Imaging, SS-2011
Mohammad Dawood
Optical flow
Brightness consistency constraint
With Taylor expansion
V : Flow (Motion)
28
Medical Imaging, SS-2011
Mohammad Dawood
29
Medical Imaging, SS-2011
Mohammad Dawood
Optical flow
Lucas Kanade Algorithm: Assume locally constant flow
=>
30
Medical Imaging, SS-2011
Mohammad Dawood
Optical flow
Horn Schunck Algorithm: Assume globally smooth flow
31
Medical Imaging, SS-2011
Mohammad Dawood
Optical flow
Bruhn’s Non-linear Algorithm
32
Medical Imaging, SS-2011
Mohammad Dawood
Visit
23.05.2011 14:00
EIMITechnologiehof, Mendelstr. 11
48149 Münster
www.uni-muenster.de/eimi