Flexible Algorithms for Image Registration

Foredragsholder: Prof Jan Modersitzki

Institute of Mathematics and Image Computing

University of Lübeck, Germany

Foredraget vil bli holdt på engelsk. Sammendrag (engelsk) følger.

Abstract:

Image registration is one of the challenging problems in image processing. Given two images taken for example at di erent times, from di erent devices or perspectives, the goal is to determine a reasonable transformation, such that a transformed version of one of the images is similar to the second one. There is a large number of applications demanding for registration. Areas range from art, astronomy, astro-physics, biology, chemistry, criminology, genetics, physics, or basically any area involving imaging techniques. More speci c examples include remote sensing (generating a global picture from di erent partial views), security (comparing current images with a data base), robotics (tracking of objects), and in particular medicine, where computer-aided intervention, fusion of different modalities, intervention and treatment planning, monitoring of diseases, motion correction, or radiation therapy demand for registration. As motion provides one of the key challenges in medical imaging, it is no surprise that an overwhelming number of publications address this issue;see, e.g. [1-5] and references therein.

In this talk, we discuss the registration problem from di erent perspectives by showing a variety of medical applications. We then present a uni ed variational approach that can be used to model, compare and design registration tools. A key design element is modularity which enables an adaption to the needs of a particular applications. In this framework, the objective is to minimize a joined functional, which is the weighted sum of a data  tting term and a regularization. Several modeling aspects as well as numerical issues will be discussed and illustrated. We also exemplarily discuss how to use the FAIR toolbox, which is freely downloadable from the website http://www.siam.org/books/fa06/.

If time permits, we also discuss the incorporation of additional information such for example rigidity of bones and the compensation of intensity variations.

References

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

[2] J. Hajnal, D. Hawkes, and D. Hill, Medical Image Registration, CRC, 2001.

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

[4] , FAIR : Flexible Algorithms for Image Registration, SIAM, 2009.

[5] O. Scherzer, Mathematical Models for Registration and Applications to Medical Imaging, Springer, New York, 2006.

[6] T. S. Yoo, Insight into Images : Principles and Practice for Segmentation, Registration, and Image Analysis, AK Peters Ltd, 2004.