A novel approach to pattern recognition in real-time arrhythmia detection

A description is given of an attempt to find an effective method to detect an arrhythmia in a real-time processing environment. In place of the conventional techniques of template-matching feature extraction used in automated arrhythmia detection, autocorrelation techniques are used. This approach makes use of the fast Hartley transform to calculate the autospectrum between two waveforms. By comparing only 10 samples of the discrete autospectrum, it is possible to compare effectively 256 samples which scan the whole of an ECG (electrocardiogram) cycle.<>