Application of calculation of correlation dimension for description of the various arrhythmias in ECG signal

Recently the theory of nonlinear dynamics has been used for describing and analysing various biomedical signals. From signals based on heart activity only the signal of the heart rate has been analysed in more details. The electrocardiogram recordings from the MIT-BIH arrhythmia database have been preliminary analysed in the framework of recent methods of nonlinear dynamics. In the authors' research, they have dealt with the reconstruction of the two-dimensional and three-dimensional attractors of ECG signals by the Taken's method of delays (Fojt, O. and Holcik, J.; 1995) and with the estimation of the correlation dimension of ECG signals by the Grassberg-Procaccia algorithm (Fojt, O. and Holcik, J.; 1995). The authors obtained almost always a fractional value of the correlation dimension confirming the chaotic nature of the electrical heart activity measured from ECG. However, the computing of the correlation dimension is often influenced by the choice of some working parameters. The main purpose of this work is to gain insights on the influence of such factors on results and to demonstrate some examples of heart arrhythmias and their description.