gms | German Medical Science

Question: Can we characterize the spatiotemporal complexity of atrial arrhythmias based on clinical signals?

Method Used: A 3D biophysical model of the atria was developed, which simulates the propagation of the electrical impulse based on the Courtemanche membrane kinetics. Eighteen different episodes of atrial fibrillation (AF) were generated, as well as an episode of atrial flutter and normal rhythm. The episodes differ by the arrhythmogenic substrate that was created in order to make the model vulnerable to AF. The substrates were specified by introducing heterogeneities in action potential duration or a stable focal source (with a fixed firing rate). The time course of the equivalent dipole was computed, derived from the equivalent double layer expression of the sources. The smallest eigenvalue S of the covariance matrix of the 3 components of the dipole with normalized magnitude was taken as a measure of spatial complexity of the distribution. Its value is near 0 for distributions concentrated around a great circle or a point and near 1 for uniform ones. To test the applicability of this approach for clinical data, the same measure was computed from the VCG derived by means of a dedicated set of transfer coefficients from the 12-lead ECG of AF patients, after QRST cancellation.

Results: The value of S was found to be < 0.1 during normal rhythm and flutter, and in the range 0.1-0.5 in macro-reentrant AF episodes. More complex forms of AF were associated with larger values of S (0.5-0.8). The maximal value was reached for a disorganized atrial activity generated by 3-6 wavelets widely dispersed over the atria. The presence of a focal activity tended to decrease the value of S. In clinical signals, the observed values of S were 0.4+/-0.13 (range: 0.13-0.78) during AF (n=70) and < 0.2 during flutter (n=2) and < 0.1 normal rhythm (n=2).

Conclusion: The results suggest that information about the spatial complexity of AF can be derived from the VCG.