Design and evaluation of time-alignment methods for averaging low level electrocardiographic data.

Seven different time-alignment method used in averaging of cardiac signals recorded in the presence of noise were evaluated and compared. The degree of precision the seven methods provided was evaluated by the ability of each to enhance large amplitude signals such as the QRS complex as well as the finer details such as low level potentials, commonly seen and known to be effective in characterizing arrhythmia susceptibility in post-infarction patients. The seven alignment methods under study were single lead voltage threshold, single lead minimum derivative, single frame template, tow frame template, three frame template, single lead correlation, and multilead correlation. Both simulated and measured electrocardiographic data were used assess the advantages disadvantages, and accuracies of the methods. Simulated data were used a permit control variables such as signal-to-noise ratio, types of additive noise cycle length variation, and to provide a check of the alignment and averaging programs. Real data were used to demonstrate alignment performance. All methods aligned without error when applied to simulated data that did not contain noise. The two correlation methods produced the smallest alignment errors and correlation using multiple leads was shown to produce the least alignment error when applied to both types of data under various levels and types of additive noise. Further investigation using multiple leads for correlation was made resulting in the study, produced the least error in comparison to the other methods when applied to various data types and noise condition used in this study.