Constructing activation time series from local and regional unipolar waveform analysis

A new method is proposed to improve automatic detection of activation over protracted sequences of unipolar signals recorded at the surface of the heart. A set of four scalar and vectorial functions is calculated involving either a single local signal or an increasing number of neighboring signals. A primary sequence of activation times is detected using the time derivative of the local signal. Templates are then constructed to represent the temporal variation of the four functions within a 10–20 ms interval centered on these primary events. Each template is translated along its corresponding function and a correlation coefficient is calculated; a threshold is set beyond which new activations are detected. If new detections are added, they are included in the templates and the entire process is repeated in an iterative fashion.