Epicardial potential distributions calculated from body surface measurements using multiple torso models

This thesis describes the development of a system for the calculation of epicardial potentials from measured body surface electrocardiographic data, using highly detailed computerised models of the torso. To provide data so that the system can be evaluated in a clinical situation, a body surface mapping system which can be used easily on sick patients has been developed. The equipment is located in a trolley which can be wheeled to a patient's bedside. A 'jacket' of electrodes is used to allow, simple and rapid electrode placement. In order to calculate the forward and inverse transfer matrices required to calculate epicardial potentials, an automated system for modelling three dimensional volume conductors has been developed. This system has been validated by comparison of the modelled potential distribution due to a dipole located in a sphere with the known analytical solution. Initial calculations performed on a highly detailed inhomogeneous model of the human torso show that, by using a regularisation method, physiologically plausible epicardial potentials can be calculated from measured body surface potentials. Four torso models with identical torso shape but varying internal conductivities, and seven other models with varying torso shape were used to explore the effects of varying conductivity and geometry on the forward and inverse calculations. It was found that omission of the lungs or bony structures caused variations in both the forward and inverse calculations, with the lungs having a major effect. Variations in external torso geometry also caused sizeable changes in both the forward and inverse calculations.