Method for the computation of an accurate zero reference for ECG signals

When the instantaneous magnitude of an electrical signal is to be determined, a reference has to be known against which the signal value should be measured. In electrocardiology, this reference is commonly derived from the isoelectrical period in the ECG. However, this period is not a clearly defined part of the ECG, and its use can be problematic. The signal level obtained from it is often not sufficiently unambiguous and in the case of a high heart rate this period is too short to be used or not present at all. In the paper a method is developed to compute a reference value, independent of the isoelectrical period and of heart rate. The method is derived from a method published before, which we developed for the separation of the surface His-Purkinje signal from the P-wave in the case of overlap. In the present application of the method, two successive heartbeats are separated in an analogous manner, revealing the signal level at zero heart activity. In applications described in the paper, this signal level is used as zero reference in nonlinear techniques. The method proved to work well and accurately.