The segmented-beat modulation method for ECG estimation.

Abstract

Electrocardiographic (ECG) tracings corrupted by noise with frequency components in the ECG frequency band, may result useless unless appropriately processed. The estimation of the clean ECG from such recordings, however, is quite challenging; being linear filtering inappropriate. In the common situations in which the R peaks are detectable, template-based techniques have been proposed to estimate the ECG by a template-beat concatenation. However, such techniques have the major limit of not being able to reproduce physiological heart-rate and morphological variability. Thus, the aim of the present study was to propose the segmented-beat modulation method (SBMM) as the technique that overcomes such limit. The SBMM is an improved template-based technique that provides good-quality estimations of ECG tracings characterized by some heart-rate and morphological variability. It segments the template ECG beat into QRS and TUP segments and then, before concatenation, it applies a modulation/demodulation process to the TUP-segment so that the estimated-beat duration and morphology adjust to those of the corresponding original-beat. To test its performance, the SBMM was applied to 19 ECG tracings from normal subjects. There were no errors in estimating the R peak location, and the errors in the QRS and TUP segments were low (≤65 μV and ≤30 μV, respectively), with the former ones being significantly higher than the latter ones. Eventually, TUP errors tended to increase with increasing heart-rate variability (correlation coefficient: 0.59, P<;10(-2)). In conclusion, the new SBMM proved to be a useful tool for providing good-quality ECG estimations of tracings characterized by heart-rate and morphological variability.