Author + information
- Received January 22, 2019
- Revision received September 17, 2019
- Accepted September 19, 2019
- Published online January 8, 2020.
- Rui Shi, MD, PhDa,b,
- Paras Parikh, PhDc,
- Zhong Chen, MBBS, PhDb,
- Nathan Angel, PhDc,
- Mark Norman, MDb,
- Wajid Hussain, MDb,
- Charlie Butcher, MBBSb,
- Shouvik Haldar, MDb,
- David G. Jones, MDb,
- Omar Riad, MDb,
- Vias Markides, MDb and
- Tom Wong, MDb,∗ ()
- aDepartment of Cardiovascular Medicine, The First Affiliated Hospital of Xi’an Jiaotong University, Xi’an, China
- bHeart Rhythm Centre, The Royal Brompton and Harefield National Health Service Foundation Trust, National Heart and Lung Institute, Imperial College London, United Kingdom
- cAcutus Medical, Carlsbad, California
- ↵∗Address for correspondence:
Dr. Tom Wong, Heart Rhythm Centre, The Royal Brompton and Harefield National Health Service Foundation Trust, National Heart and Lung Institute, Imperial College London, Sydney Street, London SW3 6NP, United Kingdom.
Objectives This study sought to validate the accuracy of noncontact electrograms against contact electrograms in the left atrium during sinus rhythm (SR) and atrial fibrillation (AF).
Background Noncontact mapping offers the opportunity to assess global cardiac activation in the chamber of interest. A novel noncontact mapping system, which records intracardiac voltage to derive cellular charge sources (dipole density), allows real-time mapping of AF to guide ablation.
Methods Noncontact and contact unipolar electrogram pairs were recorded simultaneously from multiple locations. Morphology correlation and timing difference of reconstructed electrograms obtained from a noncontact catheter were compared with those from contact electrograms obtained from a contact catheter at the same endocardial locations.
Results A total of 796 electrogram pairs in SR and 969 electrogram pairs in AF were compared from 20 patients with persistent AF. The median morphology correlation and timing difference (ms) in SR was 0.85 (interquartile range [IQR]: 0.71, 0.94) and 6.4 ms (IQR: 2.6, 17.1); in AF was 0.79 (IQR: 0.69, 0.88) and 14.4 ms (IQR: 6.7, 26.2), respectively. The correlation was stronger and the timing difference was less when the radial distance (r) from the noncontact catheter center to the endocardium was ≤ 40 versus > 40 mm; 0.87 (IQR: 0.72, 0.94) versus 0.73 (IQR: 0.56, 0.88) and 5.7 ms (IQR: 2.6, 15.4) versus 15.1 ms (IQR: 4.1, 27.7); p < 0.01 when in SR; 0.81 (IQR: 0.69, 0.89) versus 0.67 (IQR: 0.45, 0.82) and 12.3 ms (IQR: 5.9, 21.8) versus 28.3 ms (IQR: 16.2, 36.0); p < 0.01 when in AF.
Conclusions This novel noncontact dipole density mapping system provides comparable reconstructed atrial electrogram measurements in SR or AF in human left atrium when the anatomical site of interest is ≤ 40 mm from the mapping catheter.
- atrial fibrillation
- contact mapping
- noncontact dipole density mapping
- unipolar signal
Understanding patient-specific non–pulmonary vein mechanisms in persistent atrial fibrillation (AF) is essential to establish effective ablation strategies that improve clinical outcomes. Sequential mapping of AF is challenging using conventional contact catheters given the continuously changing patterns of activation. The instantaneous and global nature of noncontact mapping may help to elucidate the underlying patient-specific mechanisms contributing to the perpetuation of AF (1,2).
A novel noncontact dipole density mapping system (Acutus Medical, Carlsbad, California) was designed to overcome some limitations of existing technologies and deliver improved temporal and spatial resolution (3). A double layer of dipole density (coulombs [C]/cm) forms during propagation that embodies the wave front and generates the cardiac potential field (volts). The fundamental differences between voltage and dipole density lie in both the averaging effect of “spatial summation” and in the volume of space occupied by each (4). Theoretically, dipole density–based mapping provides a more localized portrayal of activation patterns than voltage-based mapping does, with less far-field interference. However, validation of dipole density mapping in the human left atrium (LA) has not been previously reported.
The purpose of this study was to validate the accuracy of unipolar reconstructed electrograms by this dipole density–based noncontact mapping system against conventional contact electrograms obtained from a conventional circular catheter during sinus rhythm (SR) and AF in the human LA.
This study included 20 consecutive patients with persistent AF who underwent an electrophysiologic study followed by radiofrequency catheter ablation guided by noncontact dipole density mapping. The written consent forms were signed by all patients before the procedure, and the study was approved by the national ethics committee (National Health Service Health Research Authority).
Noncontact dipole density mapping and validation design
The noncontact mapping catheter (AcQMap; Acutus Medical) has a 25-mm diameter spheroid-shape formed by 6 splines, with each containing 8 electrodes and 8 ultrasound transducers (Central Illustration, Panel A). The system reconstructs 3-dimensional endocardial anatomy from reflection points marked by ultrasound transducers. An impedance-based localization system tracks the position of electrodes, integrates ultrasound distances, and registers the acquired surface mesh within a unified coordinate system. Dipole density is inversely derived from the raw, unipolar cavitary electrograms sensed by the noncontact electrodes. Reconstructed electrograms are “forward-calculated” from the derived distribution of dipole density. In contrast to interpolation between points of measured data, inverse solutions predict data based on an appropriate model that fundamentally describes the behavior of the data. Ultimately, comparable models with low measurement noise can accurately predict values farther away from the locus of measurement points than approximate models with high measurement noise. The presence of measurement noise requires the application of regularization to prevent large-scale multiplication of noise into an erroneous prediction. These challenges and limitations apply to the present method under evaluation and are accordingly contained within the results.
An electric dipole consists of 2 oppositely charged particles separated by a small distance. When a myocardial cell is stimulated, ionic currents flow through membrane ion channels and a small dipolar imbalance in charge emerges in the adjoining extracellular medium. As propagation proceeds, the combined activation of cells forms a macroscopic double layer of dipole density, in units of C/cm, that physically embodies the wave front and generates the cardiac potential field, in units of volts (5). It is precisely this potential field that contact and noncontact electrodes sense. Charge density, itself, cannot be physically measured. Only its force (voltage) can be measured and, thus, dipole density must be inversely derived by a mathematical algorithm.
Four key stages of methodology affect the accuracy of this dipole density–based noncontact mapping system: 1) inputs to the inverse solution, including the noncontact, voltage-based electrograms and geometric coordinates of the electrodes and surface mesh; 2) mathematics of the inverse solution that derive dipole density; 3) post-processing of dipole density to reconstruct the unipolar, voltage-based electrograms on the anatomical surface; and 4) detection of local activation time for the display of wave front propagation from either dipole density or voltage. Methods of activation detection (methodology 4) and associated errors are known and common to all mapping technologies and are not examined in this study. However, the first 3 methodologies are unique to this mapping system and therefore constitute our scope of examination (see detailed explanation in the Online Appendix, Methods 1 and 2). Because dipole density cannot be physically measured, it was not directly examined in this in vivo study. Accordingly, this validation was limited to examination of the reconstructed unipolar voltage. Specifically, the accuracy of this mapping system was assessed using cross-correlation to compare morphology and timing of reconstructed voltage-based electrograms, post-processed from the dipole density–based inverse solution, to contact electrograms, measured from a conventional circular catheter, at the same locations across the LA endocardial surface.
Electrophysiology study and electroanatomic mapping procedures
Clinical application of noncontact dipole density mapping has been described in detail previously (4). Specific to this study, a standard decapolar contact catheter was placed in the coronary sinus and a quadripolar catheter was placed in the inferior vena cava below the level of the diaphragm, with 1 electrode used as an electrical reference for unipolar recording. The LA was accessed via double trans-septal punctures. The noncontact catheter and 20-pole circular catheter (Inquiry Optima; Irvine Biomedical, Inc., Abbott Laboratories, Irvine, California) with 2-mm electrodes, 15- to 25-mm variable loop radii were introduced into the LA via a 16-F steerable trans-septal sheath (AcQGuide, Acutus Medical) for recording and mapping. Heparin was administered by intravenous bolus to maintain an activated clotting time ≥350 s to prevent thrombus formation. The noncontact catheter was fully deployed over a 0.032-inch guidewire, after having been positioned in the middle of the LA for both acquisition of endocardial anatomy and recording cardiac activation (Central Illustration, Panel B).
By maneuvering and rotating the deployed noncontact catheter, ultrasonic reflections were detected and processed to reconstruct the atrial chamber anatomy. With the deployed noncontact catheter positioned in the center of the LA, a conventional circular catheter was navigated to 7 pre-specified regions in the LA for simultaneous recording of contact and noncontact unipolar electrograms on the septum; roof; floor; left atrial appendage (LAA); and the lateral, anterior, and posterior walls (Central Illustration, Panel C). The noncontact mapping catheter in its fully deployed state was positioned and rotated within the atrial chamber without direct contact with the atrial endocardium when creating the anatomy of the LA. The noncontact mapping catheter was positioned generally in the center of the chamber when performing the electrical mapping. There is no requirement to hold the AcQMap catheter in a fixed position, as the AcQMap system continuously tracks the position of the catheter spline electrodes and inputs the associated coordinates for each instant of inverse calculation. Comparisons of contact and reconstructed electrograms were thereby made from multiple recordings at those 7 pre-specified regions during SR and AF.
All recorded segments of AF and SR were captured for a duration of 30 s and were reviewed and analyzed off-line. SR was restored either by catheter ablation or external cardioversion, if required. The AcQMap methodology applies an algorithm to determine local activation time and organizes activation time as a moving window of isochronal timing, called a propagation history map (5).
Validation and accuracy methods
Data including the coordinates of surface mesh points, localized contact and noncontact catheter electrodes, and all associated unipolar electrograms were exported for off-line analyses using custom MATLAB scripts (MathWorks, Natick, Massachusetts). For AF recordings, 300- to 600-ms segments containing multiple cycles were selected for analysis, whereas single-atrial deflections were selected for SR recordings. All segments were manually chosen to ensure complete exclusion of QRS waves and T waves.
Correspondence between noncontact reconstructed and contact electrograms was evaluated using the mean-subtracted cross-correlation (see detailed explanation in the Online Appendix, Methods 3). In addition, the root-mean-square value of each of the selected segments were calculated for the electrogram pairs obtained during SR. Amplitude was not calculated and compared for AF segments due to the dramatic variation in amplitude over time. In contrast, SR is an inherently regular rhythm with consistent and repeatable amplitude measures that are appropriate for analysis. Metrics comparing noncontact to contact electrograms have been previously described (6,7).
Each mapping catheter electrode was indexed to a specific geodesic vertex on the anatomy. Cross-correlation was then calculated at each surface-mesh vertex residing within 5 mm from the index vertex. The maximum correlation value found within this 5-mm radius was assigned as the morphology match value for that contact and noncontact electrogram pairs. This radius accounted for inverse-solution and electrode-localization errors and ensured the most relevant morphologic assessment. Factors including contact catheter distance from the anatomy, stability of mapping catheter localization, and baseline noise were examined to achieve appropriate comparisons of electrical data.
The data was pooled across all patients as a single rhythm group, for both SR and AF. Analysis of morphology was done by using the standard cross-correlation metric. Pearson correlation is calculated across the time series for each sample offset, and a set of Pearson correlation values are thereby accrued. Cross-correlation is defined as the maximum correlation value obtained within the accrued set of values. Timing difference is defined as the time difference associated with sample offset required to achieve cross-correlation. These 2 metrics were calculated, individually, for every location of all contact electrodes and were accumulated into a population of independent morphology and timing measures. Continuous and normally distributed data were presented as mean ± SD, and an independent Student’s t-test was used to test differences between 2 groups. Normally distributed data were expressed as median (interquartile range [IQR]) and compared using the non-parametric Mann-Whitney U test. Statistical calculations were made in SPSS (version 22.0; IBM Corp., Armonk, New York). The results were considered significant for a p value < 0.05.
Twenty patients (mean age 64 ± 12 years, 14 male) with persistent AF (mean AF duration 9 ± 6 months) were studied. The mean LA parasternal long-axis diameter was 43 ± 5 mm and left ventricular ejection fraction was 49 ± 14% (see detailed patients’ characteristics in Online Table 1).
Validation of reconstructed electrograms in SR
A total of 796 SR electrogram pairs were compared, with results summarized in Table 1. Median morphology cross-correlation and timing difference was 0.85 (IQR: 0.71, 0.94) and 6.4 ms (IQR: 2.6, 17.1), respectively. Morphology correlation and timing difference as a function of distance from the center of the noncontact catheter during SR are summarized in Figure 1. Both the morphology correlation and timing difference of electrogram pairs improved when the radial distance (r) between the endocardial comparison site and the center of the noncontact catheter was ≤40 mm, with 91.5% of the sites located within this range. As shown in Figure 1, median morphology correlation was 0.87 (IQR: 0.72, 0.94) versus 0.73 (IQR: 0.56, 0.88); p < 0.01 and timing difference was 5.7 ms (IQR: 2.6, 15.4) versus 15.1 ms (IQR: 4.1, 27.7); p < 0.01 for r ≤ 40 mm versus r > 40 mm, respectively. Electrogram pairs at equatorial sites had a significantly greater correlation than at polar sites: median: 0.85 (IQR: 0.73, 0.95) versus 0.81 (IQR: 0.67, 0.92); p = 0.01. However, polar sites were significantly farther from the center of the noncontact catheter: median: 32.0 mm (IQR: 24.7, 40.2) versus 28.0 mm (IQR: 21, 33.6); p < 0.01.
The amplitude of contact electrograms in SR increased as the contact catheter was moved closer to the endocardial surface and farther in radial distance from the center of the noncontact catheter: median: 0.22 mV (IQR: 0.14, 0.40) versus 0.35 mV (IQR: 0.22, 0.46); p < 0.01 for r ≤ 40 mm versus r > 40 mm, respectively. Conversely, the amplitude of corresponding non-contact-reconstructed electrograms decreased: median: 0.15 mV (IQR: 0.10, 0.22) versus 0.10 mV (IQR: 0.08, 0.14); p < 0.01 for r ≤ 40 mm versus r > 40 mm, respectively. Endocardial sites for which the amplitudes of electrogram pairs were closely matched were located closer to the center of the noncontact catheter: median: 21.2 mm (IQR: 16.2, 30.2) versus 28.4 mm (IQR: 23.2, 35.2); p < 0.01 for root-mean-square amplitude ratio ≥ 0.9 versus < 0.9, respectively. These sites also had a correspondingly better morphology correlation: median: 0.94 (IQR: 0.83, 0.98) versus. 0.80 (IQR: 0.67, 0.90); p < 0.01. As a general observation, non-contact-reconstructed electrograms with the lowest amplitudes were observed at sites furthest from the center of the noncontact catheter (Figure 1).
Validation of reconstructed electrograms in AF
A total of 969 AF electrogram pairs were compared, with results summarized in Table 1. Median morphology correlation and timing difference were 0.79 (IQR: 0.69, 0.88) and 14.4 ms (IQR: 6.7, 26.2), respectively. Central Illustration, Panel D illustrates an example of a dataset comparing 10 pairs of non-contact-reconstructed electrograms (red waveform) to the corresponding contact electrograms (blue waveform) on the lateral wall of LA.
Similar to observations in SR, morphology correlation and timing difference of AF electrogram pairs improved when the radial distance between the endocardial comparison site and the center of the noncontact catheter was ≤40 mm, with 85% of the sites located within this range (Table 1). As shown in Figure 2, median morphology correlation was 0.81 (IQR: 0.69, 0.89) versus 0.67 (IQR: 0.45, 0.82); p < 0.01 and median timing difference was 12.3 ms (IQR: 5.9, 21.8) versus 28.3 ms (IQR: 16.2, 36.0); p < 0.01 for r ≤ 40 mm versus r > 40 mm, respectively. Electrogram pairs at equatorial and polar sites had the essentially the same morphology correlation: median: 0.80 (IQR: 0.69, 0.88) versus 0.81 (IQR: 0.69, 0.89); p = 0.28. However, timing differences for equatorial and polar sites were significantly different: median: 16.6 ms (IQR: 8.2, 30.1) versus 13.3 ms (IQR: 6.0, 24.0); p < 0.01.
Morphology correlation and timing difference at the specific regions of the LA for AF are shown in Figure 3. The regions with the highest median morphology correlation were observed on the roof: 0.87 (IQR: 0.77, 0.91), and posterior wall: 0.86 (IQR: 0.76, 0.90). The region with the lowest morphology correlation was within and near the LAA ostium: median: 0.70 (IQR: 0.59, 0.81). Of note, the LAA region was also the furthest from the center of the noncontact catheter: median: 39 mm (IQR: 31, 51).
This is the first study to compare dipole density–based reconstructed unipolar electrograms with contact electrograms in the human LA. This signifies the first quantitative step in validating dipole density–based mapping using physically comparable measurements. Morphology correlation and timing difference values were significantly improved at radial distances ≤40 mm, with most of the comparison sites located within this range (more than 85% in this evaluation).
Previous validation study of noncontact mapping
The global, noncontact mapping mode of the EnSite System (Abbott Laboratories, St. Paul, Minnesota) established a new concept in the history of 3-dimensional mapping, as shown by previous studies (8,9). Earley et al. (6) compared the contact unipolar electrograms with noncontact “virtual” unipolar electrograms by using the noncontact EnSite Array catheter in the human LA during SR and AF. LA electrograms correlated well (0.81) during SR. However, the cross-correlation values in AF dropped from 0.72 to 0.59 when ventricular far-field signals were excluded from correlation analysis by inducing temporary atrioventricular block with adenosine. This implied that the common-mode ventricular component of the unipolar electrogram may have contributed a favorable bias in the assessment of the EnSite system, with an associated overestimation of its accuracy in mapping AF (6). Differences exist between the present study and the data reported by Earley et al. (6). In their assessment of AF, the QRS complex was included in the signal morphology over a duration of approximately 7 s. Inclusion of the QRS complex artificially improves the timing difference, as the far-field QRS complex is inherently well-aligned in time across all measurements and reconstructions, and this feature dominates the determination of timing difference. In contrast, the present assessment methodology did not include the QRS complexes. Instead, a multitude of shorter-duration signals were compared for time segments between, and not including, QRS complexes. A stronger cross-correlation (0.85 and 0.79) was shown in SR and AF, respectively. This supports the view that dipole density–based mapping can be used to detect endocardial activity of AF with increased accuracy (3).
An inherent limitation of noncontact mapping is that the morphology correlation is dependent on the distance between the noncontact mapping catheter and the endocardial wall. In the validation studies using the EnSite Array catheter, Earley et al. (6) and Lin et al. (10) reported a significant decrease in morphology correlation for endocardial points >40 mm and >38 mm from the center of the array, respectively. Those studies also indicated greater variation in the morphology correlation and timing difference for specific regions of the LA, such as the more distant LAA and mitral-annular sites (6,10). In the present validation study, the distance from the center of the noncontact catheter to the endocardial wall also imposed a similar limitation on the accuracy and precision of voltage-based measurement. However, the results showed good accuracy and precision (correlation 0.79 in AF) with ventricular complexes excluded from the calculation of cross-correlation.
Independent from morphology correlation, electrogram timing is also an important metric of evaluation (10). In this study, timing difference gradually increased with increasing distance between the center of the noncontact catheter and the endocardial comparison sites. Timing accuracy is known to be affected by complex geometry, such as the shape of the appendage and venous antra (11). Timing errors are imposed in different ways on both the non-contact-reconstructed and contact electrograms from the complex geometric shape of LA anatomy and from propagating wave fronts having a “sharp” curvature. Delay caused by catheter tissue contact presents another known dimension of variation in comparing noncontact and contact electrogram timing. Such geometric- and tissue-pressure–related factors were not specifically evaluated in this study. Nonetheless, such factors are inherent within these reported results. Electrogram comparison in this study methodology does not rely on phase for measurement of activation times. Phase calculation is not performed for any purpose in the AcQMap algorithms or in the comparison analysis. Consequently, phase discontinuity does not manifest as a limitation in the present study.
There are 2 steps to manipulate the noncontact mapping catheter during the procedure: 1) to create the atrial anatomy; and 2) to acquire the electrical data.
1. The mapping catheter was positioned within the atrial chamber in its fully deployed state and was then rotated and translated within the chamber to close proximity but not in direct contact with the endocardium to reconstruct the atrial anatomy.
2. The catheter was then positioned generally in the center of the chamber for performing the electrical mapping. It is not necessary to “hold” the catheter in a fixed position. The system localizes the electrodes positions in real time; and their coordinates are applied in the inverse algorithm for every instant of calculation. The signals associated with electrodes located within 5 mm of the endocardial surface were automatically rejected by the algorithm to maintain the integrity of noncontact mapping processing.
This novel noncontact dipole density mapping system harnesses the advantage of reconstructing an anatomy generated by high–spatial resolution ultrasound and deriving dipole density that provides more localized maps of cardiac conduction than voltage (5). Non–pulmonary vein triggers and specific substrate conditions have become decisively relevant in treating persistent AF. Technologies are evolving toward localization of targets for ablation outside the pulmonary veins (1,12–15). Contact mapping technologies have limitations of either low–spatial resolution “global” or high-resolution sequential (regional) acquisition of data points. Technologies using phase algorithms may lead to a false-positive bias in detecting rotational activities and focal sources (16). Dipole density mapping addresses some of the limitations of conventional mapping and may contribute to uncovering the mechanisms of AF that remain to be fully elucidated.
We have shown in this study that noncontact electrograms collected by dipole density–based mapping are reliable in morphology correlation and timing difference within a radial distance of 40 mm from the center of the noncontact catheter. The resulting morphology correlation between noncontact and contact electrograms was the least at the LAA, which was usually the furthest distance from the center of the catheter splines than the rest of the atrial anatomy. Thus, the noncontact electrograms and cardiac conduction maps by dipole density mapping during AF must be interpreted with caution if anatomical areas are over 40 mm from the center of the noncontact catheter, such as in the region of the LAA. In practical terms, it is advisable for the novel noncontact catheter to be placed generally in the middle of the LA chamber during mapping. In doing so, it is likely that the majority of endocardial aspects of the LA will be within optimal range.
First, dipole (charge) density is an electrical entity that cannot, itself, be directly measured by contact mapping, although charges and voltages are inherently linked. Accordingly, this study was limited to comparing post-processed voltage-based electrograms to corresponding contact electrograms at the same location on the chamber wall. The reconstructed noncontact electrograms therefore have the associated inherent limitations from its derivation methodology and the inherent reproduction of far-field blending. Second, validation analysis was limited to segments within the R-R intervals to avoid interference from ventricular electrograms. The associated short duration of compared sample sets would impose some limitations on evaluation of activation stability in both the time domain and in phase space. Consequently, evaluation of stability was not within the scope of this validation. Third, the study did not assess the impact of anatomical geometry on measurement locations that were either within or adjacent to the direct field-of-view of the noncontact catheter on the accuracy of data collection. The data presented in the present study includes comparison of signals recorded and reconstructed during AF. We propose that the data presented would be predictive of a future study that could be contemplated using bipolar signals, in addition to unipolar signals, to validate the accuracy of complex fractionated atrial electrograms.
This novel noncontact dipole density mapping system provides comparable reconstructed atrial electrogram measurements in SR or AF in human LA when the anatomical site of interest is ≤40 mm from the mapping catheter.
COMPETENCY IN MEDICAL KNOWLEDGE: This novel noncontact dipole density mapping system harnesses the advantage of reconstructing an anatomy generated by high–spatial resolution ultrasound and deriving dipole density that provides more localized maps of cardiac conduction than voltage. Noncontact electrograms collected by dipole density–based mapping are reliable in morphology correlation and timing difference within a radial distance of 40 mm from the center of the noncontact catheter. Thus, the noncontact electrograms and cardiac conduction maps by dipole density mapping during AF must be interpreted with caution if anatomical areas are over 40 mm from the center of the noncontact catheter.
TRANSLATIONAL OUTLOOK: Non–pulmonary vein triggers and specific substrate conditions have become decisively relevant in treating persistent AF. Contact mapping technologies have limitations of either low–spatial resolution “global” or high-resolution sequential (regional) acquisition of data points. Technologies using phase algorithms may lead to a false-positive bias in detecting rotational activities and focal sources. Dipole density mapping fundamentally addresses these limitations and may contribute to uncovering the mechanisms of AF which has thus far been a challenging quest.
The authors wish to acknowledge and thank Xinwei Shi, PhD, and Min Zhu, PhD (algorithm scientists at Acutus Medical, Inc.) for their contributions to the analytic methodology.
Dr. Shi received a fellowship scholarship from the China Scholarship Council (No. 201606285148) paid to the institution. Dr. Markides has a consulting agreement with Biosense Webster. Dr. Wong received support from National Institute for Health Research. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- atrial fibrillation
- interquartile range
- left atrium
- left atrial appendage
- sinus rhythm
- Received January 22, 2019.
- Revision received September 17, 2019.
- Accepted September 19, 2019.
- 2019 The Authors
- Haissaguerre M.,
- Hocini M.,
- Denis A.,
- et al.
- Shi R.,
- Norman M.,
- Chen Z.,
- Wong T.
- Grace A.,
- Willems S.,
- Meyer C.,
- et al.
- Earley M.J.,
- Abrams D.J.,
- Sporton S.C.,
- Schilling R.J.
- Bear L.R.,
- LeGrice I.J.,
- Sands G.B.,
- et al.
- Markides V.,
- Schilling R.J.,
- Ho S.Y.,
- Chow A.W.,
- Davies D.W.,
- Peters N.S.
- Steinhaus B.M.
- Cuculich P.S.,
- Wang Y.,
- Lindsay B.D.,
- et al.
- Pandit S.V.,
- Jalife J.
- Narayan S.M.,
- Baykaner T.,
- Clopton P.,
- et al.