Signal processing methods to improve high resolution ECG signal averaging

Coherent signal averaging as a method to enhance signal to noise ratio (SNR) relies on the determination of a fiducial point for temporal alignment of signals. Various alignment methods have been proposed but they produce alignment errors when the signal is corrupted with noise, resulting in a low-pass filtering effect on the averaged signal. In high resolution ECG, alignment errors can affect late potential detection. In this paper, a method is presented to reduce the alignment error using cross correlation and pre-processing of the ECG. Simulations using real ECG signals and additive gaussian noise indicate a 30% to 50% reduction on the alignment error standard deviation when compared to previous studies.