Electrocardiogram characterization using wavelet analysis

The electrocardiograph (ECG) is a graphical representation of the forces generated during cardiac activity, and is an essential tool for the diagnosis of cardiac abnormalities. An automatic ECG analyzer will provide a cardiologist with a tool allowing faster and more accurate diagnosis. The analysis consists of the measurement of the amplitudes, durations and morphologies of the P, QRS and T waves. This paper deals with the measure of QRS duration, R spikes detection (arrhythmia), and the starting and vanishing time of the T wave. A comparison of different methods based on the derivatives and filtering for the extraction of ECG characteristics is presented. We show that wavelet analysis gives better results than classical methods, and enables a finer characterization of the parameters