Automated feature extraction of ECG signal by position-index searching method

An automated ECG signal analysis is a progressively growing field to the researchers in recent years as it contains several features in itself which are clinically important to diagnose different cardiovascular diseases. This paper introduces a simple but reliable algorithm for automatic detection of different features in an ECG signal by position-indices searching method (PISM). In the first step, the digitized ECG signals is denoised by moving average method. A differential threshold value is adopted to detect the indices of the positions of the R-peaks of the signal. Then by omitting the baseline wandering, the amplitudes of R-peaks are detected. In each cycle of normal ECG, all the features in ECG signal, namely P, Q, S & T-peaks, have their own amplitudes and positions with respect to their corresponding R-peak positions. In next step by taking the R-peak as pivot point, the amplitudes and position-indices of these features are detected by PISM. This algorithm is tested on a number of normal ECG records from PTB Database of variant ages and sex. The suggested algorithm shows very high sensitivity and positive predictivity.