Reconstruction of Equivalent Human Heart Dipole Parameters for the Torso Model Represented by Elliptical Conducting Cylinder of a Finite Length

In this paper we solve the inverse problem of electrocardiography by using nonlinear least-square optimization algorithm. At first we test this algorithm on the modeled signals of the dipole with known parameters: for the dipole with one projection of the moment vector and for the dipole with random arbitrary moment vector. At second we test this algorithm with real electrocardiographic signals obtained from the human. In the case of real signals we obtain two time-spatial plots - the tracks (trajectories of dipole movement) and hodographs (trajectories of dipole moment change). We compare hodographs with well known vectorcardiographic loops of Ernest Frank's lead system.