Automatic detection and analysis of cardiac arrhythmias in long ECG-recordings

The paper is devoted to research and development of new ECG processing methods, based on Multiscale Correlation Analysis approach, in particular - the development of new cardiac arrhythmias detection method. It is shown that in case of ECG, where a signal has a form of repeated wave pulses, the instruments previously developed by authors and called the analytical spectra technique are highly effective. Computational algorithm based on this technique not only accurately estimates the heart rate, but can also detect its sharp changes. The detection of such changes together with subsequent analysis of jump shape enables the automatic recognition of premature beats on ECG. The paper provides first results received in this direction. The results of PVC detection for long ECG recordings from standard MIT-BIH NSRDB are briefly summarized in the appropriate section and shortly characterized in the conclusions.