How accurately can the method of fundamental solutions solve the inverse problem of electrocardiology?

Abstract

This study presents a detailed comparison between the Method of Fundamental Solutions (MFS) approach to solving the inverse problem of electrocardiology and a more conventional boundary element method (BEM) approach. Synthetic data were created to simulate the heart surface potential distribution during the time course of normal and ectopic heart beats. Both measurement and geometry noise were added to the data and the inverse problem was solved via both methods. Under these conditions several regularisation parameter determination methods were compared, with the Robust Generalised Cross-Validation (RGCV) method consistently performing better than any other method for both MFS and BEM approaches. The MFS approach to solving the inverse problem of electrocardiology can sometimes yield more accurate results than the BEM approach, especially when the regularisation parameter is determined by RGCV, but BEM is generally superior.