Robust detection of ECG waves

In this paper we present an overview of a new method for ECG wave detection and labelling. It is based on a wavelet transform of the ECG signal, but expands upon previous approaches by including both multi-lead and multi-scale treatment of the input signal. This unique property allows for a very robust operation with relatively simple processing. Our ECG wave detector is designed to be robust against changes in sampling rate, recording noise and missing leads. Our current implementation is developed and tested on standard 12 lead ECGs sampled with a sampling rate of 500 Hz and 10 second segments. We work in the wavelet domain by applying dyadic discrete wavelet transforms to each lead separately. This representation is merged into a single signal, which is processed uniformly and independently of the initial number of leads.