Spatial characteristics of atrial fibrillation using the surface ECG
Publikation/Tidskrift/Serie: Computers in Cardiology
The present study investigates spatial properties of atrial fibrillation (AF) by analyzing VCG loops synthesized from 12-lead ECGs. During intervals with ventricular activity, the atrial signal is extracted using spatiotemporal QRST cancellation, thus making continuous f-wave analysis possible. Spatial properties are characterized through joint analysis of successive fixed-length signal segments, either of 1-s length or related to the typical length of an f-wave. The spatial properties are expressed in terms of loop orientation, i.e., angles of azimuth and elevation, as well as in morphologic terms such as planarity and planar geometry of the loops and related temporal variability. Several parameters have been suggested to characterize the degree of AF organization from electrogram analysis, but only a few parameters for noninvasive, ECG-based characterization of which AF frequency is the most studied. This parameter is related to atrial refractoriness and reflects AF organization. In the present study, the significance of spatial (loop morphology) parameters is studied in relation to AF frequency, hypothesizing that more organized AF, being expressed by a lower frequency, is associated with decreased variability in loop morphology. A total of 26 patients with chronic AF were analyzed with respect to spatial properties, using 60-s ECG recordings. For the database AF frequency ranged from 3.9 to 7.7 Hz, with a mean of 6.4±1.0 Hz (mean±std). The results showed that loop orientation can be determined from global analysis (i.e., the entire 60-s segment was used for determining the two angles) or the mean of segment-based analysis, both types of analysis leading to similar results. The hypothesis that more organized AF is associated with decreased variability in loop morphology was to some extent confirmed for the parameters planarity and planar geometry; for 1-s segments, the correlation to AF frequency was 0.608 (p<0.001) and 0.543 (p<0.005). This finding also applied to segment lengths related to f-wave duration. In conclusion, the simpler analysis based on 1-s segments may be preferred since it does not require f-wave delineation. Global analysis and the mean of segment-based analysis (1-s segments) showed increased planarity and decreased planar geometry for lower AF frequencies, possibly related to higher similarity of successive loops in more organized AF.
- Cardiac and Cardiovascular Systems
Computers in Cardiology, 2007
- Signal Processing Group-lup-obsolete
- ISSN: 0276-6574
- ISBN: 978-1-4244-2533-4