Classification of ECG using chaotic models

Abstract

Chaotic analysis which works all the nonlinear dynamics ofa system has been shown to be useful in a variety of medical applications, particularly in cardiology. Chaotic parameters have shown potential in the identification of diseases that can be identilled [.rom bic)mcdieal signals like electrocardiogram (ECG). In this work. underlying cha'os in ECG signals has been analyzed using various non- Iinear techniques. First, the ECG signal is processed through a series of steps to extract the QRS complex. From this extracted feature, bit-to-bit interval (BBI) and instantaneous heart rate (IHR) have been calculated. Some statistical parameters like mean, standard deviation (SD) and eoeflicient of variation arc calculated hom BBI and IHR time series. The nonlinear techniques like phase space portrait and central tendency measure (CTM) have been applied to the BBI and IHR. Standard database of MIT-BIH is used as the reference data. Twenty two records normal and abnormal rhythms of this database arc analyzed where each record is of halr-hollr duration and contains around (,50.000 samples. Phase space portrait of both BBI and IHR demonstrates visible attractor with little dispersion for healthy person's ECG and a widely dispersed plot in 2-D plane lar the abnormal ECG. CTM is calculated far both BBI and IHR lar each ECG record of the database. A much higher value (0.7737 x 0.0946) of CTM ror IHR is observed for eleven patients with normal beats while that far eleven patients with abnormal rhythm is much low (0.0833 x 0.0748) and the values are signilicantly different. CTM lar BBI or the same eleven normal records is 0.6172 x 0.1472 and thaI for eleven abnormal records is 0.0478 x 0.0308. These rcsults indicate that ECG can be classilied based on chaotic modelling.