Quantitative Assessment of Synchronization during Atrial Fibrillation by Shannon Entropy Characterization of Propagation Delays M Masè 1 , F Ravelli 1,2 , L Faes 1 , R Antolini 1 , G Nollo 1,2 1 Department of Physics and INFM, University of Trento, Italy 2 ITC-irst, Trento, Italy Abstract This study introduced a new method for the quantification of the synchronization (S) and the causal verse of activation (S12) in couples of atrial electrograms recorded during atrial fibrillation (AF). The synchronization indexes S and S12 relied on the measure of the propagation delays between coupled activation times in two atrial signals and on the characterization of their dispersion by Shannon-Entropy (SE). S and S12 were validated both on simulated activation time series and endocavitary signals in patients. In simulation, S and S12 were equal to 1 for propagation of one single wavefront in a fully excitable tissue, while they decreased for reentries in partially excitable tissue (S = 0.70 ± 0.05, S12 = 0.66 ± 0.05 ) and multiple wavelet propagation (S = 0.46 ± 0.06, S12 = 0.39 ± 0.08). In patients S, was equal to 1 during atrial flutter (AFl) and decreased with increasing complexity of AF (AF1: S = 0.76 ± 0.05; AF2: S = 0.56 ± 0.06; AF3: S = 0.39 ± 0.03). Moreover S12 evidenced the preservation of a correct activation sequence during AFl and AF1 (S12 = S) and its loss during AF2 (S12 = 0.41 ± 0.12 < S) and AF3 (S12 = 0.26 ± 0.03 < S). As indirect markers of the electrophysiological properties of atrial tissue, indexes S and S12 may provide a new insight in understanding the mechanisms initiating and maintaining AF and support new clinical treatments for its interruption. 1. Introduction Although AF has been referred as a totally disorganized rhythm, a growing body of work has demonstrated that, also during this arrhythmia, the activation of the atrial chambers is affected by a certain number of factors (as anatomy, refractoriness etc), which may play an important role in limiting the randomness of propagation and giving “organization” to AF. As the interest in AF has grown in the scientific community, many methods to quantify this aspect have been proposed [1]. These followed different philosophical and methodological ways, due to the lack of an universally recognized definition of “organization”. In this study we focus on a specific aspect of organization, namely the synchronization, which concerns the relationship between the electrical activities of couples of atrial sites. To quantify this aspect we introduce two different measures of the level of synchronization: the indexes of synchronization (S) and of causal coupling (S12). S and S12 rely on a Shannon Entropy measure of the dispersion of the propagation delays between coupled activations in two atrial sites. In fact they are based on the assumption that, if two atrial sites are repeatedly interested by the passage of the same stable activation-wave (i.e. are synchronized), the consecutive propagation delays between coupled activations recorded in the sites should have a small dispersion. On the contrary, when the sites are activated by the same but unstable wave or by different wavelets [2], the values of the delays should present a higher variability. The two indexes S and S12 were validated and compared on cellular automaton (CA) simulated activation time series and on endocavitary signals in patients. 2. Methods 2.1. Index of synchronization S The level of synchronization in the electrical activity of two atrial sites was defined by a Shannon Entropy- characterization of the distribution of the propagation delays between coupled activations in the two sites. Once the activation time series had been estimated for both signals by a morphology-based algorithm [3], the propagation delays were defined as the absolute value of the differences between coupled activations in the two signals. Couples were obtained by associating each activation in one series with the closest in the other series. Then the propagation delay values were organized in a histogram and their spreading was characterized by a measure of Shannon Entropy (SE): ∑ = ⋅ - = N i i p i p SE 1 ) ( ln ) ( (1) 0276-6547/04 $20.00 © 2004 IEEE 257 Computers in Cardiology 2004;31:257-260.