Improving the signal-to-noise ratio of electrocardiogram (ECG) measurements facilitates cardiac beat detection in a human or animal patient. ECG signals measured either on the surface of the skin or subcutaneously from pairs of leads may be combined to calculate a differential signal. The measured signal may also be averaged to produce a second estimate. A point-by-point product of the differential signal and the averaged signal is generated if the sample pair has the same polarity. The product signals feature enhanced signal components and reduced noise components, thus improving the signal-to-noise ratio of the respective input channels. Product signals are then subjected to peak detection through a conventional auto-sensing approach. Preliminary event detection results from the conditional product signals may then be aggregated, and final sense markers for ECG beat detection may be generated by means of a voting algorithm.
