Nonlinearity characterization and entropy analysis of intracardiac atrial electrogram signals

The main objective of this paper is to identify evidences of nonlinear components in persistent atrial fibrillation electrograms (persAF). Firstly, an exploratory data analysis using linear approaches (autocorrelation and spectral analysis) was performed to assess the behaviour of the AF electrograms. Secondly, a nonlinear characterization using surrogate data analysis was performed with a complementary return map comparison. Finally, Approximate Entropy (ApEn) was considered as an analysis tool of system complexity to measure disorganization over time for different AF behaviours. From our results, (1) we identified strong evidence of nonlinear components negating the null hypothesis, (2) lower ApEn values were associated with organized and periodic AF activations and higher entropy values were associated with the increase of electrogram complexity and (3) ApEn response of longer window segments showed a general AF behaviour while shorter windows (1s and 2s) help identifying dynamical atrium electrical changes.