Delineation of cardiac late potentials using ensemble correlation

Beat averaging reduces the influence of noise in electrocardiogram (ECG) signals by taking advantage of the fact that noise is essentially uncorrelated from beat to beat. This fact is further developed in this study for determining the endpoint of late potentials in individual leads. The ensemble of beats is first split into subaverages and then highpass filtered. The crosscorrelation between the two filtered beat subaverages will remain high as long as late potential activity is present. The endpoint is estimated by using a simple threshold procedure on the resulting correlation function. For patients with late potentials, it was found that the endpoint occurred considerably later with the new method when compared to standard techniques. In normal ECGs, the endpoint remained essentially unchanged