The inverse problem in electrocardiography: solutions in terms of epicardial potentials.

The objective of the inverse problem in electrocardiography is to recover noninvasively regional information about intracardiac electrical events from electrical measurements on the body surface. The choice of epicardial potentials as the solution to the inverse problem is motivated by the availability of a unique epicardial potential solution for each body surface potential distribution, by the ability to verify experimentally the inverse-recovered epicardial potentials, by the proven relationship between epicardial potentials and the details of intracardiac regional events, and by the possibility of using the inverse solution as a supplement or possible replacement to clinical epicardial potential mapping prior to surgical intervention. Although, in principle, the epicardial potential distribution can be recovered from the body surface potential distribution, the inverse problem in terms of potentials is ill-posed, and naive attempts to reconstruct the epicardial potentials result in incorrect solutions which are highly oscillatory. Large deviations from the actual solution may result from inaccuracy of the data measurement, incomplete knowledge of the potential data over the entire torso, and inaccurate description of the inhomogeneous torso volume conductor. This review begins with a mathematical and qualitative description of the inverse problem in terms of epicardial potentials. The ill-posed nature of the problem is demonstrated using a theoretical boundary value problem. Effects of inaccuracies in the body surface potential data (stability estimates) are introduced, and a sensitivity analysis of geometrical and inhomogeneity parameters is presented using an analytical eccentric spheres model. Various computational methods for relating epicardial to body surface potentials, i.e., the computation of the forward transfer matrix, are described and compared. The need for regularization of the inverse recovery of epicardial potentials, resulting from the need to invert the ill-conditioned transfer matrix, is demonstrated. Several regularization techniques are compared in terms of their performance regarding noise in the data and inaccuracies in geometry and inhomogeneities. Finally, several existing, regularized inverse procedures that compute epicardial potentials from measured body surface potential data are introduced and compared. The review concludes with a section that points toward future directions for improving the quality of the inverse-reconstructed epicardial potentials. Future directions for the use of the inverse problem to obtain epicardial potential distributions noninvasively in both experimental animals and patients in a clinical se