While LDDMM methods and algorithms have proven their worth when comparing macroscopic differences between organs, it is ill-suited to generate a heartbeat, which requires a torsion in the muscle that does not naturally appear through shape matching. In this talk, we show various possible models to generate a heartbeat from a relaxed state to a contracted state while in the framework of LDDMMs, using for example constraints or artificially generated fibers, and their success (or lack of success) in modeling a heartbeat while keeping computations to a manageable level.

This presentation is part of Minisymposium “MS63 - Geometric methods for shape analysis with applications to biomedical imaging and computational anatomy”
organized by: Martin Bauer (Florida State University) , Nicolas Charon (Johns Hopkins University) .

Authors:

Sylvain Arguillere (Institut Camille Jordan)

Keywords:

image registration, nonlinear optimization