Registration Algorithms for Time-Lapse Sequences and Multimodality Imaging

P. Thévenaz, M. Unser

Autumn Meeting of the Swiss Society of Pharmacology and Toxicology: Imaging Technologies and Their Applications in Pharmacology and Toxicology (ITAPT'03), Basel BS, Swiss Confederation, October 27-28, 2003.

The accurate alignment of two images or volumes with respect to one another is a recognized need in a wide variety of applications. A mundane example is the stitching of a mosaic of images to create a panorama of your holidays scenery; but another use for the technique is to build a larger field of view from a series of images obtained through a microscope. Other examples involve the detection of minute changes, be it from a series of images obtained by an eye-fundus camera that monitors the progression of a bolus of contrast agent through the retina, or from a series of volumetric fMRI scans of the brain of a subject who alternatively performs—and rests through—a task.

Registration is also necessary when multimodal data have to be brought into a common system of coordinates. A typical example would involve a patient who undergoes an MRI scan to detect a brain tumor, a PET scan to assess its functional damage, and a CT scan to plan for its surgical removal. As the patient takes a slightly different posture in each scanner, the data from MRI, CT, and PET cannot be overlaid. Only after successful registration will the physician be able to relate the location of the various participating tissues (brain, tumor, skull).

At the Swiss Federal Institute of Technology Lausanne, we have developed three solutions to the registration problem. The first, freely distributed as an ImageJ plugin, addresses the need for monomodal 2D registration and includes the optional use of landmarks as guidance in the difficult cases; a streamlined interface facilitates the operation of this plugin. The second, freely distributed as an ANSI-C routine, addresses the need for monomodal 3D registration [1]; this routine can be tuned in many ways to achieve fine control. The third solution addresses the need for multimodal 3D registration [2].

References

1. P. Thévenaz, U.E. Ruttimann, M. Unser, "A Pyramid Approach to Subpixel Registration Based on Intensity," IEEE Transactions on Image Processing, vol. 7, no. 1, pp. 27-41, January 1998.

2. P. Thévenaz, M. Unser, "Optimization of Mutual Information for Multiresolution Image Registration," IEEE Transactions on Image Processing, vol. 9, no. 12, pp. 2083-2099, December 2000.

@INPROCEEDINGS(http://bigwww.epfl.ch/publications/thevenaz0303.html,
AUTHOR="Th{\'{e}}venaz, P. and Unser, M.",
TITLE="Registration Algorithms for Time-Lapse Sequences and
	Multimodality Imaging",
BOOKTITLE="Autumn Meeting of the Swiss Society of Pharmacology and
	Toxicology: {I}maging Technologies and Their Applications in
	Pharmacology and Toxicology ({ITAPT'03})",
YEAR="2003",
editor="",
volume="",
series="",
pages="",
address="Basel BS, Swiss Confederation",
month="October 27-28,",
organization="",
publisher="",
note="")