Tutorial on Flexible Algorithms for Image Registration

September 18th 2011, MICCAI 2011 Toronto

Organisers

Jan Modersitzki, Lars Ruthotto and Fabian Gigengack

Abstract

The tutorial provides an introduction to image registration and the design of efficient and stable state-of-the-art algorithms. It is based on [3] and the publicly available FAIR toolbox which enables easy access to even complex schemes. The tutorial enables participants to use the software, explore existing registration techniques, and to design and develop new solutions.

The tutorial also covers the integration of new problem specific features like mass-preservation and hyperelasticity. Based on important medical applications, particular demands on registration schemes are identified and the tuning of algorithms is demonstrated. In particular, solutions for motion correction in cardiac PET and susceptibility correction in EPI-MRI are addressed.

Outline

  • Sept. 18th,  8:30-10:30  Introduction to image registration (Jan Modersitzki)
  • Sept. 18th, 10:30-10:45 Coffee break 
  • Sept. 18th, 10:45-11:30 Motion Correction in cardiac PET (Fabian Gigengack)
  • Sept. 18th, 11:30-12:00 Susceptibility correction in EPI-MRI (Lars Ruthotto)

Application 1: Motion correction in cardiac PET

Respiratory and cardiac motion leads to image degradation in positron emission tomography (PET) studies of the human heart. A mass-preserving (respectively density or radioactivity-preserving) registration approach was implemented as an add-on to the FAIR toolbox. It will be shown how the new transformation model and the corresponding objective function are designed and how, e.g., the existing interpolation schemes, multilevel strategy and optimization tools can be used.

Application 2: Susceptibility correction in EPI-MRI

In fast magnetic resonance imaging (MRI) sequences such as echo-planar imaging (EPI), artifacts occur due to field-inhomogeneities. It will be demonstrated how a variational image registration approach to correct those EPI distortions can be implemented as an application of the FAIR toolbox. Based on this and the preceding example, general guidelines for including own applications will be given.

Objectives

The tutorial provides educational material for training of students, clinicians and new researchers in the field of image registration. It covers the wide range from theoretical background including well-posedness and regularization via efficient and fast implementations to timely and practical relevant medical applications.

Audience

The targeted participants are clinicians, computer scientists, engineers, and all others with interest or demand for registration.

Literature

[1] A.A.Goshtasby: 2-D and 3-D Image Registration. Wiley Press, New York, 2005.

[2] J.Hajnal, D.Hawkes, D.Hilli: Medical Image Registration. CRC Press, 2001.

[3] J. Modersitzki: FAIR: Flexible Algorithms for Image Registration. SIAM, Philadelphia, 2009.
[bib] 

[4] J. Modersitzki: Numerical Methods for Image Registration. Oxford University Press, New York, 2004.
[bib]