8C080 - Algorithms for biomedical informatics and image analysis - 2013

This course consists of two parts:

- Algorithms for Image Analysis (tutor: prof. Bart M. ter Haar Romeny). Info: see below.
- Algorithms for Informatics (tutor: prof. Peter A.J. Hilbers). Info: see here.

Algorithms for image analysis

Algorithms for image analysis aim to answer complex questions about images. Such questions arise for example in the field of computer aided diagnosis, and in the preprocessing for three-dimensional visualization.

In the course we focus on two main areas:

  1. Image filtering (convolution, Fourier series) for image enhancement / improvement and feature detection.
  2. Image registration (matching).

Course and Guided Self Study material:

Date Topics Study material Location
Tu 05-02-2013

10:45-12:30
Regular course

Introduction
The concept of convolution
Powerpoint: Introduction (12 MB)
MMA 9 notebook: The concept of convolution
PAV M23
We 06-02-2013

10:45-12:30
Guided self study

Exercise with Mathematica 9.

NB: Install Mathematica 9 beforehand on your laptop!

Download MMA9 from the TU/e campus software site.
First remove previous Mathematica versions (7 or 8).

MMA 9 screencasts: screencast   Other screencasts

MMA 9 notebook: short introduction to Mathematica 9

Rehearsal: 8ZZ16 Introduction to Mathematica
(tutorial for version 8 is the same as for version 9)

MMA 9 notebook: Template Matching

GEM-Z 3A08
Erik Bekkers

GEM-Z 3A10
Bart ter Haar Romeny
Fr 08-02-2013

13:45-15:30
Guided self study

Exercise:
Template matching
Continue with MMA 9 notebook: Template Matching

Answer

For those who managed to find the letters:
The Point-spread Function

Answer

GEM-Z 3A12
Birgit Plantinga

GEM-Z 3A13
Bart ter Haar Romeny
Tu 19-02-2013

10:45-12:30
Regular course

Fourier series,
Fourier transform,
convolution theorem
Powerpoint: Introduction to Fourier analysis
PDF: Fourier Series (course notes, 12 MB)
MMA 9 notebook: Fourier analysis (2 MB)
MMA 9 notebook: Dither removal
Java applet: Fourier series
PAV M23
We 20-02-2013

10:45-12:30
Guided self study
 
Exercise:
Fourier series,
Fourier transform,
convolution theorem (part I)
MMA notebook: Exercises with Fourier Series

Answer

GEM-Z 3A08
Birgit Plantinga

GEM-Z 3A10
Bart ter Haar Romeny
Fr 22-02-2013

13:45-15:30
Guided self study
 
Exercise:
Fourier series,
Fourier transform,
convolution theorem (part II)
See We 20-02-2013. GEM-Z 3A12
Birgit Plantinga

GEM-Z 3A13
Paul Bloembergen
Tu 26-02-2013

10:45-12:30
Regular course

Image interpolation


Geometrical transformations

PPT: Interpolation and Transformations
MMA notebook: Image interpolation
PDF: Interpolation
MMA notebook: Geometrical transformations
MMA notebook: 2D rotation-translation
MMA notebook: 3D point warpings
QTVR movie: image stitching

Movie: Goose - Britisch mode

PAV M23
We 27-02-2013

10:45-12:30
Guided self study

Exercise:
Geometric transformations
MMA notebook: exercises with geometrical transformations

Answer

GEM-Z 3A08
Erik Bekkers

GEM-Z 3A10
Paul Bloembergen
Fr 01-03-2013

13:45-15:30
Guided self study

Exercise:
Image interpolation
MMA notebook: exercises with image interpolations

Answer

GEM-Z 3A12
Paul Bloembergen

GEM-Z 3A13
Erik Bekkers
Tu 5-03-2013

10:45-12:30
Regular course

Medical image registration Powerpoint: Image Registration (11 MB)
PDF: Medical Image Registration (16 MB, *)
PDF: Distance measures
MMA notebook: Mutual Information 01
MMA notebook: Mutual Information 02
MMA ntebook: Steepest Ascent in Mathematica
WMV movie: Rotation and its cost function
PAV M23
We 6-03-2013

10:45-12:30
Guided self study

The Image Processing part ends here.

Exercise:
Medical image registration
MMA notebook: Exercises with image matching

Answer

GEM-Z 3A08
Birgit Plantinga

GEM-Z 3A10
Erik Bekkers
Fr 8-03-2013 No BZ!    

*) Chapter 3: "Registration Methodology: Concepts and Algorithms" from "Medical Image Registration", edited by Joseph V. Hajnal, Derek L.G. Hill, David J. Hawkes, CRC Press, 2001. ISBN 0-8493-0064-9.
From this chapter everything, but not the sections 3.4.1.2, 3.4.2, 3.4.2.1, 3.4.2.2,.3.4.2.3, and 3.4.9.
Section 3.4.7 is for reading only, is not part of the exam.

Material for the exam:

Example exam 01   Answers

Example exam 02   Answers

Example exam 2011   Answers

Exam 2012    Answers

These example exams are just an indication for the type of questions. The real exam is of course different, and possibly more complex.

1.       Course notes: Fourier Series 01:
 

a.       5. Complex numbers

b.      10. Fourier series

                                                   i.      The formulas for the calculation of the Fourier coefficients must be know by heart.
 

2.       Course notes: Fourier Series 02:
 

a.       Tasks 10

b.      11. Tabels: read only, and see if you understand them. If a formula is needed during the exam, it will be supplied..

3.       Medical Image Registration:

a.       Chapter 3: "Registration Methodology: Concepts and Algorithms" uit het boek "Medical Image Registration", Joseph V. Hajnal, Derek L.G. Hill, David J. Hawkes (editors), CRC Press, 2001. ISBN 0-8493-0064-9.
From this chapter all, except the sections 3.4.1.2, 3.4.2, 3.4.2.1, 3.4.2.2,.3.4.2.3, en 3.4.9.
Section 3.4.7: read only, is no part of the exam.

4.       Interactive course notes in the form of a Mathematica noteboek (play with the formulas, to understand better what happens):

a.       The concept of convolution

b.      Fourier analysis

                                                   i.      1. Fourier Series

                                                 ii.      2. Examples of Fourier series

                                                iii.      3. Convolution is a product in the Fourier domain: the convolution theorem.

                                               iv.      4. A linear system

                                                 v.      5. The response on a spike function: the Point Spread Function (PSF)

                                               vi.      6. Spatial frequencies

                                              vii.      7. Blurring in the Fourier domain

                                            viii.      8. Mathematical morphology

                                               ix.      9. Smart filters (alleen doorlezen)

c.       Image interpolation

                                                   i.      Lineaire interpolatie moet je uit je hoofd kunnen doen, in 2D en 3D.

d.      Dither removal

                                                   i.      Understand what happens here.

e.      Geometrical transformations

                                                   i.      Learn and be able to explain: Affine transformations in 2D and 3D, rotations in 2D, around the origin, and rotations around an arbitrary point.

                                                 ii.     Understand, but you do not have to implement this: deformations with an arbitrary vectorfield.

f.        3D point warpings

g.       Point set matching (Procrustes algorithm)

h.    Image registration (matching)

                    i. Know which costs functions exist to minimize for matching
               ii. Know Correlation, Least Squared Distance, Mutual Information, Joint Probability Histogram

5.       The powerpoint files are illustrations of the material above. Read them. You must be able yo explain intuitively what is:

a.       the principle of the gradient descent algorithm

Not part of the exam material:

Last updated: 30 January 2013.