A proposed method for the inverse problem in electrocardiology

Abstract

The inverse problem in electrocardiography is considered. A method is proposed in which the cardiac electrical generator is represented by a set of dipoles, fixed in location and direction in order to reflect the known features of myocardial excitation, but variable in strength. A crucial innovation is that since the dipole directions have been so chosen, the dipole strengths must be constrained nonnegative. Surface potentials are measured in vivo and the dipole strengths inferred. In this process, torso models with a varying degree of realism are used. An 11-dipole set is used and potentials are measured at 126 surface locations. For a particular normal subject, the effect of various variables, such as the torso modeling assumptions, on the dipole strengths is investigated. Condensed results are given for twelve normal subjects and two patients.