Averaging multiple EKG waveforms

Individual EKG waveforms (i.e. a single cycle from a particluar lead) can be extracted from epochs containing many such waveforms. These single waveforms can then be shifted and synchronized about a common feature present in all of them, and can be averaged together on a point by point basis. This averaging can be done in an attempt to reduce the noise content of the signal, to look for differences between individual waveforms, or to look for differences between the average and individual waveforms (e.g. "late potentials").

Here is an example of extracting and averaging 4 single cycle waveforms from a 4 second epoch of a V3 EKG signal recorded previously:

Each of the extracted waveforms is synchronized about the R peak of the trace, which has been arbitrarily placed 1/4 of a second (64 points) into the cycle. Currently this extraction and synchronization is perfomed manually, but eventually it will be performed automatically by the software from multiple epochs. The entire V3 epoch is also shown for comparison. Waveforms which cross the ends of the epoch are simply padded with zeros:

The averaging process appears to have little effect on this particular signal, since there is a low noise content, and each individual trace is very similar to one another. However, signals contaminated by additional noise components, or which vary from waveform to waveform, may benefit more from this technique.

Here are plots showing a comparison between the average signal (in red) and one of its individual waveforms (in blue). Also shown is the difference between these two waveforms. Finally, the magnitude of the spectrum of the average (in red) is shown compared to the spectrum of the difference (in blue). The difference spectrum is also shown at a larger scale for convenience.


©Copyright Sky Coyote, 2001-2002.