Session P86.3

Organization Tracking of Long-Term Atrial Fibrillation Recordings: Differences between Paroxysmal and Persistent Episodes

R Alcaraz*, F Sandberg, L Sörnmo, JJ Rieta

Universidad Politecnica de Valencia
Valencia, Spain

Several authors have reported that strong correlation exists between atrial fibrillation (AF) organization, defined as how repetitive is the AF signal pattern, and the number of wavelets wandering the atrial tissue. In this work, a method is proposed for the purpose of evaluating non-invasively AF organization such that paroxysmal and persistent long-term AF recordings can be discriminated. This is crucial in patients without AF history since the presence of AF, even during periods as short as 24 hours, has been shown to alter atrial electrophysiological properties to favor AF maintenance. Thus, early intervention could maximize the effectiveness of AF treatment.
In order to provide robust and reliable results, long-term variations of AF organization were assessed in the present study. The dataset consisted of 13 24h-Holter recordings with persistent AF and 11 with paroxysmal AF (episode lengths between 4 minutes and 4.5 hours). To evaluate AF organization, the atrial activity (AA) signal was first obtained using the spatiotemporal QRST cancellation technique. Next, the dominant atrial frequency (DAF) of the AA was tracked with a recently published method based on a hidden Markov model. This technique provides a reliable short-term DAF frequency trend, even in noisy ECG recordings. Finally, the AA main atrial wave (MAW) was obtained by selective filtering centered on the DAF, thus providing a time series suitable for computation of the sample entropy (SampEn). The SampEn-based MAW regularity analysis has previously demonstrated to be a robust estimator of AF organization.
The results show that episodes of paroxysmal AF were consistently associated with lower SampEn than episodes with persistent AF. In numerical terms, the overall SampEn mean value for paroxysmal AF was 0.0693±0.0147, whereas it was 0.1056 ± 0.0146 for persistent AF (p=4.064•10-6). As a consequence, a SampEn threshold of 0.087 could be established to discriminate between paroxysmal and persistent AF. Moreover, the minute-by-minute tracking of AF organization showed that paroxysmal AF was 94.2% of the time below the threshold, whereas persistent AF was 88.6% of the time above, thus making it possible to identify long-term recordings of patients without AF history.

(Abstract Control Number: 35)