Section: Research Program
Simplified models and inverse problems
The medical and clinical exploration of the cardiac electric signals is based on accurate reconstruction of the patterns of propagation of the action potential. The correct detection of these complex patterns by noninvasive electrical imaging techniques has to be developed. This problem involves solving inverse problems that cannot be addressed with the more compex models. We want both to develop simple and fast models of the propagation of cardiac action potentials and improve the solutions to the inverse problems found in cardiac electrical imaging techniques.
The cardiac inverse problem consists in finding the cardiac activation maps or, more generally, the whole cardiac electrical activity, from highdensity body surface electrocardiograms. It is a new and a powerful diagnosis technique, which success would be considered as a breakthrough. Although widely studied recently, it remains a challenge for the scientific community. In many cases the quality of reconstructed electrical potential is not adequate. The methods used consist in solving the Laplace equation on the volume delimited by the body surface and the epicardial surface. Our aim is to

study in depth the dependance of this inverse problem on inhomogeneities in the torso, conductivity values, the geometry, electrode positions, etc., and

improve the solution to the inverse problem by using new regularization strategies, factorization of boundary value problems, and the theory of optimal control.
Of course we will use our models as a basis to regularize these inverse problems. We will consider the following strategies:

using complete propagation models in the inverse problem, like the bidomain equations, for instance in order to localize electrical sources;

constructing families of reducedorder models using e.g. statistical learning techniques, which would accurately represent some families of wellidentified pathologies; and

constructing simple models of the propagation of the activation front, based on eikonal or levelset equations, but which would incorporate the representation of complex activation patterns.
Additionaly, we will need to develop numerical techniques dedicated to our simplified eikonal/levelset equations.