Atrial flutter and atrial fibrillation have different generating mechanisms in the atria. However, they are often cross-classified from the surface ECG. Nonlinear analysis has recently been applied to electrograms, and atrial arrhythmia has shown evidence that indicates the possibility of deterministic chaos. In this study, we applied methods from the theory of nonlinear dynamics to characterize electrograms of atrial flutter and atrial fibrillation in humans. For typical flutter, nonlinear parameters were relatively smaller, and they presented higher values when in atrial fibrillation. In atypical flutter, the magnitudes of these nonlinear parameters were between those of typical flutter and fibrillation.
These parameters exhibited a significant differentiation, allowing the classification of these arrhythmias. By using the neural network classification, a desirable result was obtained. Therefore, nonlinear analysis provides us an advantageous technique to discriminate among typical flutter, atypical flutter and fibrillation electrograms from surface ECG.