Average beat subtraction before compression reduces the data rate necessary for ambulatory ECG storage by approximately 100 b/s.
The theoretical rate-distortion bounds for compression of ambulatory electrocardiograms (ECGs) are presented. These bounds estimate the minimum data rate for coding the ECG signal with a given mean-square-error distortion relative to ECGs recorded with a 100-Hz bandwidth. The rate-distortion bounds are calculated from the average power spectrum of an arrhythmia database. The calculations indicate that average coding distortions on the order of 15 mu V RMS are required for storage of 24 h of two-channel ECG in 4 MB. The ECG can be stored with average distortions on the order of 11 mu V when average beats are subtracted before compression. Average beat subtraction reduces the data rate necessary for storage by approximately 100 b/s regardless of distortion level.>
Hamilton et al. (Mon,) studied this question.