NA-MIC/Projects/Structural/Optimal transport for Registration

Objective:

We want to develop new elastic registration methods for brain imaging. The idea is to use optimal transport as the similarity metric underlying the registration procedure. In order to use this technique on 3D manifolds, they must first be warped to the 2D plane using conformal mapping techniques. Additionally, we would like to allow users to be able to specify anatomical landmarks in the form of artificial slits placed in the segmented brain surfaces. During the registration these slits would be forced to register to one another thus guiding the process and producing more accurate results.

Progress:

• This techniques has been implemented for standard 2D images and simple surfaces in [1] and [2] along with application to medical images.
• Code has been developed to accomplish the warping of the brain from a complex 3D surface containing several holes onto the plane.
• Optimal transport has been used to register heart [3] and vessel imagery [4]. These applications required surfaces to be more complex in that they contained holes and other topological challenges such as branches. As a result, they needed more complex flattening procedures.

Ongoing:

• Further development of algorithms to enable highly complex surfaces with embedded landmark information to be registered using optimal transport.

Key Investigators:

• Steve Haker - Harvard
• Shawn Lankton - Georgia Tech
• Lei Zhu - Georgia Tech
• Allen Tannenbaum - Georgia Tech
• Ron Kikinis - Harvard

References:

• [1] Haker S, Tannenbaum A, Kikinis R. Mass Preserving Mappings and Image Registration. Proc MICCAI 2001, LCNS 2208; p 120-127
• [2] Haker S, Zhu L, Tannenbaum A, Angenent S. Optimal Mass Transport for Registration and Warping. IJCV, 60(3),225-240,2004; p 225-240
• [3] Zhu L, Haker S, Tannenbaum A. Mass Preserving Registration for Heart MR Images. Proc MICCAI 2005, LCNS 3750; p 147-154
• [4] Zhu L, Haker S, Tannenbaum A. Area-Preserving Mappings for the Visualization of Medical Structures. Proc MICCAI 2003, LCNS 2879; p 277-284

Links:

• Stony Brook Algorithms