Quantitative relation between chaotic features of surface electrocardiogram and Intracardiac Electrogram

The aim of this research is to find relations between chaotic features of the Intracardiac Electrogram signal and the surface ECG based on Partial Least Square (PLS) Regression. Intracardiac Atrial Fibrillation database from Physionet (IAFDB) is selected database for this paper, and Correlation dimension (CD), Largest Lyapunov Exponent (LLE) and Spatial Feeling Index (SFI) are considered features. Coefficients of a linear formula between the chaotic features of ECG and Electrogram were computed for each region (bipolar electrodes) by PLS Regression method. Formulas acquired for estimating CD of Electrogram using those of surface ECG in AF patients was efficient, while efforts to obtain relations for LLF and SFI did not yield significant results. Moreover, it led to an important observation that the closer bipolar electrodes were to the heart apex, the greater coefficients of lead II would be. Results of this study propose a way to eliminate the risk of invasive procedure of Electrogram recording, reduce psychological stress of patients and reduce the cost.