Lempel-Ziv complexity analysis for the evaluation of atrial fibrillation organization
Abásolo, D, Alcaraz, R, Rieta, JJ and Hornero, R (2011) Lempel-Ziv complexity analysis for the evaluation of atrial fibrillation organization In: 8th IASTED International Conference on Biomedical Engineering, 2011-02-16 - 2011-02-18, Innsbruck, Austria.
Available under License : See the attached licence file.
The Lempel-Ziv (LZ) complexity is a non-linear time series analysis metric that reflects the arising rate of new patterns along with the sequence. Thus, it captures its temporal sequence and, quite conveniently, it can be computed with short data segments. In the present work, a detailed analysis on LZ complexity is presented within the context of atrial fibrillation (AF) organization estimation. As the analysed time series depend on the original sampling rate (fs), we evaluated the relationship between LZ complexity and fs. Furthermore, different implementations of LZ complexity were tested. Our results show the usefulness of LZ complexity to estimate AF organization and suggest that the signals from a terminating paroxysmal AF group are more organized (i.e. less complex) than those from the non-terminating paroxysmal AF group. However, the diagnostic accuracy was not as high as that obtained with sample entropy (SampEn), another non-linear metric, with the same database in a previous study (92% vs. 96%). Nevertheless, the LZ complexity analysis of AF organization with sampling frequencies higher than 2048 Hz, or even its combination with SampEn or other non-linear metrics, might improve the prediction of spontaneous AF termination.
|Item Type:||Conference or Workshop Item (Paper)|
|Additional Information:||From the proceedings of the 8th IASTED International Conference on Biomedical Engineering http://www.iasted.org/conferences/home-723.html|
|Divisions:||Faculty of Engineering and Physical Sciences > Mechanical Engineering Sciences|
|Depositing User:||Symplectic Elements|
|Date Deposited:||19 Nov 2012 14:17|
|Last Modified:||23 Sep 2013 19:36|
Actions (login required)
Downloads per month over past year