Body surface potential mapping (BSPM) data obtained duringendocardial stimulation at multiple ventricular pacing sites show abroad spectrum of potential distributions. In this study, BSPM sequencesare analysed using a neural network approach based on self-organisationthat provides a noninvasive estimation of the site of origin ofstimulated ventricular activation. The Self-Organizing Map (SOM) networkused in this study is arranged as a two-dimensional lattice of neurons,each of them representing a particular distribution of body surfacepotentials. For the training of the SOM network, 123-channel BSPMrecordings were obtained from 86 endocardial pacing locations in 19patients with a previous myocardial infarction. Ventricular activationpatterns from different pacing sites are visualized as time traces onthe trained SOM. Classification of the activation patterns with respectto the endocardial pacing location is performed by Learning Vectorquantization. The localisation results are visualized on a realisticmodel of the endocardial surfaces of the right and left ventricles
展开▼
