Equivalent cardiac dipole localization from ECG data using proper orthogonal decomposition

The effort in this paper is concentrated on the localization of the equivalent heart source from the measured ECG data blending the efficiency of Finite Element Method (FEM) and the added speed of Proper Orthogonal Decomposition (POD). The chief idea is to solve a fine discretized forward eigenproblem which in turn is reduced in dimension with the aid of POD. For every forward problem iteration the voltage, only, within a small box is expressed through FEM. Voltage POD expansion is used for the remaining Human torso computational domain. To enforce continuity between the two areas a Dirichlet-to-Neumann formalism is adopted. Finally, for the inverse localization problem cost function is set up in least squares means. Both the sensitivity matrix and the calculated data sets are evaluated through a POD eigenfunction expansion. The current effort is expected to provide computational efficiency in the forward and overall inverse problem solution.