Characterization of ECG signals using multiscale approach

Electrocardiogram signals are examined for accurately diagnosing the heart abnormalities. Different signal processing techniques have been applied on these signals to interpret and detect the heart diseases. Considering the inherent self similar pattern of ECG signals as a signature for normal behaviour, the present work explores the usefulness of the associated Hurst exponent as a means for characterizing the ECG signals for their normal, anomalous behaviours. The paper applies on ECG data sets various methods that detect Hurst exponent with and without wavelet transform. Our experimental observations present the ranges of Hurst exponents that signify the presence of anomalous behaviour in ECG data.