International Journal of Bioelectromagnetism Vol. 9 No. 1 2007
Comparison of two time-domain methods to identify complex fractionated atrial electrograms for navigation during catheter ablation of electrophysiological substrate of atrial fibrillation Vaclav Kremen, Lenka Lhotska Department of Cybernetics, Czech Technical University in Prague, Czech Republic
[email protected]
analysis of AF is burdened by many methodical problems of spectral analysis [6]. This is why the software support for electroanatomical mapping system is oriented on objective description and space representation of CFAEs distribution most recently.
Abstract Complex fractionated atrial electrograms (CFAEs) represent the electrophysiological substrate of atrial fibrillation (AF). Progress in signal processing algorithms to identify CFAEs sites is crucial for the development of AF ablation strategies. We compared two methods to discriminate atrial electrograms (A-EGMs) with different degree of fractionation. First method (M-1) used approximation of an algorithm previously implemented in a commercially available electroanatomic mapping system. Second method (M-2) searched for segments of fractionation of A-EGMs and it extracted a feature used as index of fractionation. Comparison of both methods was performed using representative set of 1.5s A-EGMs (n = 113) ranked by an expert into 4 categories: 1 - organized atrial activity; 2 - mild; 3 intermediate; 4 - high degree of fractionation). Discriminative power of M-2 to detect CFAEs was superior (p = 0.02) to that provided by M-1 algorithm. Novel method of A-EGMs classification offers operator-independent definition of electrogram complexity.
2. Methods Representative set of 1500 ms length CFAEs (n=113) consisting of continuous levels of fractionation from highly organized to maximally fractionated was exported from records captured during endocardial mapping of AF sampled 977 Hz (CardiLab 7000, Prucka Inc.). The signals were ranked by an expert to four categories regarding level of fractionation: 1 - organized atrial activity; 2 - mild; 3 intermediate; 4 - high degree of fractionation. The level of fractionation was automatically quantified by two methods (M-1 and M-2). First method M-1 used as reference was based on an algorithm previously implemented in a commercially available electroanatomic mapping system [7]. M-1 calculated index of fractionation as mean of the intervals between discrete peaks of A-EGMs, which were detected using input parameters of M-1 method: peak-to-peak sensitivity, signal width and refractory period. Input parameters were optimized by exhaustive search with respect to the used data set (sensitivity 0.02mV, signal width 8ms, refractory period 14ms). Pearson’s correlation coefficient between indexes of fractionation and categories ranked by an expert was used as optimization criteria. M-2 method preprocessed signal using wavelet transform [8] filter. Coiflet wavelet of order four was used to decompose signal into 5 levels [9]. Levels were thresholded with these settings (level 1 to level 5): 0.02, 0.04, 0.008, 0.008 and 0.008. Then fractionated segments (points of interest) with electrical activation of electrophysiological substrate were detected defined by minimal amplitude (Amin), minimal segment length (Dseg) a minimal intersegments distance (Dmin). A part
1. Introduction Atrial fibrillation (AF) is a cardiac arrhythmia characterized by very rapid and uncoordinated atrial activation with a completely irregular ventricular response [1]. Radiofrequency ablation of atrial areas that triggers or sustains AF is nonfarmacological treatment available recently [2]. During AF, multiple wavefronts propagate continuously through the right and left atria, separated by anatomical and functional barriers [3]. This can be electrophysiologically manifested as hierarchical distribution of dominant frequency [4] or complex fractionated electrograms (CFAEs) [5] during endocardial mapping. Local dominant frequency
13
International Journal of Bioelectromagnetism Vol. 9 No. 1 2007
of the signal was selected as a segment in case, where the absolute value of amplitude was higher than Amin and lasted minimally Dseg. If two segments were distance more closely than Dmin, these segments were joined together and considered as one segment hereafter. Index of fractionation of the M-2 method was calculated as total number of inflexion points within all found fractionated segments in this individual A-EGM signal. Input parameters of the M-2 method were optimized in the same way as of the M-1 method (Amin = 0.003 mV; Dseg = 20 ms; Dmin = 70 ms).
where evidently no local electric activity is present and then describes fractionation in a universal way in the rest of electrogram segments. New method of A-EGM classification offers alternative definition of complex atrial electrogram. It can be easily implemented into present mapping systems and help to navigate the ablation of AF substrate in real time.
5. References 1.
3. Results 2.
Pearson’s correlation coefficients (PCC) between categories of electrograms and indexes of fractionation of both methods were considered to compare discriminative power of the methods. PCC in the M-1 method was -0.72 and PCC in the M-2 method was 0.84 for the given data set (figure 1). This difference was statistically significant with p-value p=0.02.
3.
4.
Figure 1 shows comparison of fractionation indexes of the M-1 (left) and M-2 (right) methods. r - Pearson’s correlation coefficient between indexes of fractionation and categories ranked by an expert.
5.
6.
4. Conclusions We performed direct comparison of two methods describing atrial electrograms during atrial fibrillation and we showed that fractionation index calculated by the new proposed M-2 method corresponds significantly better with categories of electrograms ranked by an expert, than index of fractionation calculated by the reference M-1 method. The M-1 method is still the only known approach to evaluating electrogram fractionation. Its exact algorithm was not published in all details, but only in the company user manual [7]. While the M-1 method is focused on searching for individual atrial complexes and thus it is very close to methods of dominant frequency classification, the newly described M-2 method primarily eliminates segments of electrograms,
7.
8.
9.
14
Fuster V, Ryden LE, Cannom DS, et al. ACC/AHA/ESC 2006 Guidelines for the management of patients with atrial fibrillation. Circulation 2006; 114:257-354. Calkins H, Brugada J, Packer DL, et al. HRS/EHRA/ECAS expert consensus statement on catheter and surgical ablation of atrial fibrillation: recommendations for personnel, policy, procedures and follow-up. Heart Rhythm 2007;4:816-61. Houben RPM, Allessie MA. Processing of intracardiac electrograms in atrial fibrillation. Diagnosis of electropathological substrate of AF. IEEE Eng Med Biol Mag 2006; 25: 4051. Sanders P, Berenfeld O, Hocini M, et al. Spectral analysis identifies sites of highfrequency activity maintaining atrial fibrillation in humans. Circulation 2005; 112:789-97. Nademanee K, McKenzie J, Kosar E, et al. A new approach for catheter ablation of atrial fibrillation: mapping of the electrophysiologic substrate. J Am Coll Cardiol 2004; 43:204453. Ng J, Kadish AH, Goldberger JJ. Effect of electrogram characteristics on the relationship of dominant frequency to atrial activation rate in atrial fibrillation. Heart Rhythm 2006; 3:1295-305. Ensite NavXTM Navigation and Visualisation Technology (Fractionation Mapping Tool Procedure Guide, Ensite 6), pp. 2-8, St. Jude Medical, 2006. Daubechies I.: Ten lectures on Wavelts, CBMS-NSF, SIAM, 61, Philadelphia, Pennsylvania, USA, 1992. Mallat, S. A Wavelet tour of Signal Processing. 2. edition. ISBN/ISSN:0-12466606-X. Academic Press. 1999.
