Methods for the detection of ECG characteristic points

The electrocardiogram (ECG) is a measurement of the electrical signals associated with the heart and is a key diagnostic tool and patient monitoring device for clinicians. This work presents and compares algorithms for per cycle temporal location of the key ECG phases using 5 methods: 2 pt and 5 pt slope, correlation, wavelet analysis and empirical mode decomposition (EMD). A new wavelet algorithm is proposed using the Daubechies DB6 wavelet and octave band decomposition followed by a series of reconstruction estimates used for temporal localization of each of the ECG phases and it is shown to detect R waves accurately (97.6% - st dev 0.048). A new EMD algorithm is proposed that uses reconstruction estimates to determine the temporal location of each of the ECG phases and it is shown to detect R waves accurately (98.5% - st dev 0.042).