Author + information
- Received August 15, 2018
- Revision received December 2, 2018
- Accepted December 4, 2018
- Published online April 15, 2019.
- Lea Melki, MSc, MPhila,∗,
- Christopher S. Grubb, BSb,∗,
- Rachel Weber, BA, RDCS, RVTa,
- Pierre Nauleau, PhDa,
- Hasan Garan, MDb,
- Elaine Wan, MDb,
- Eric S. Silver, MDc,
- Leonardo Liberman, MDc,† and
- Elisa E. Konofagou, PhDa,d,†∗ ()
- aUltrasound Elasticity Imaging Laboratory, Department of Biomedical Engineering, Columbia University, New York, New York
- bDivision of Cardiology, Department of Medicine, Columbia University Medical Center, New York, New York
- cPediatric Electrophysiology, Division of Pediatric Cardiology, Department of Pediatrics, Columbia University Medical Center, New York, New York
- dDepartment of Radiology, Columbia University Medical Center, New York, New York
- ↵∗Address for correspondence:
Dr. Elisa E. Konofagou, Biomedical Engineering, Columbia University Medical Center, 630 West 168th Street, Physicians and Surgeons 19-418, New York, New York 10032.
Objectives This study sought to demonstrate the feasibility of electromechanical wave imaging (EWI) for localization of accessory pathways (AP) prior to catheter ablation in a pediatric population.
Background Prediction of AP locations in patients with Wolff-Parkinson-White syndrome is currently based on analysis of 12-lead electrocardiography (ECG). In the pediatric population, specific algorithms have been developed to aid in localization, but these can be unreliable. EWI is a noninvasive imaging modality relying on a high frame rate ultrasound sequence capable of visualizing cardiac electromechanical activation.
Methods Pediatric patients with ventricular pre-excitation presenting for catheter ablation were imaged with EWI immediately prior to the start of the procedure. Two clinical pediatric electrophysiologists predicted the location of the AP based on ECG. Both EWI and ECG predictions were blinded to the results of catheter ablation. EWI and ECG localizations were subsequently compared with the site of successful ablation.
Results Fifteen patients were imaged with EWI. One patient was excluded for poor echocardiographic windows and the inability to image the entire ventricular myocardium. EWI correctly predicted the location of the AP in all 14 patients. ECG analysis correctly predicted 11 of 14 (78.6%) of the AP locations.
Conclusions EWI was shown to be capable of consistently localizing accessory pathways. EWI predicted the site of successful ablation more frequently than analysis of 12-lead ECG. EWI isochrones also provide anatomical visualization of ventricular pre-excitation. These findings suggest that EWI can predict AP locations, and EWI may have the potential to better inform clinical electrophysiologists prior to catheter ablation procedures.
Accessory pathways (AP) in Wolff-Parkinson-White (WPW) syndrome are commonly treated with catheter ablation (1–3). Localization of the AP prior to catheter ablation is important for pre-procedure planning. The current standard for locating the AP is the clinician’s interpretation of 12-lead ECG. However, this method is limited and localization may differ among clinicians. Many algorithms have been proposed with varying degrees of success. Specific algorithms have been developed in the pediatric population, but there is still significant room for improvement (4–9). Moreover, the reported algorithms have been less accurate than those used in adult patients, and frequently the accuracy of these algorithms in clinical practice is lower than in the originating author’s hands (9–11).
Noninvasive, more precise, and less time-consuming localization of AP could be of clinical benefit to operating electrophysiologists. Previously, there have been multiple noninvasive methods proposed for localization of AP, such as from Botvinick et al. (12). Modern electrical mapping approaches such as ECG imaging have emerged in specialized clinical settings (13). However, ECG imaging is an expensive technique that requires patient-specific models of cardiac geometry derived by computed tomography or cardiac magnetic resonance scan, potentially exposing patients to ionizing radiation or anesthesia. More recently, echocardiography strain-based methods have been explored as potential tools for the noninvasive identification of AP (14,15).
Electromechanical wave imaging (EWI) is a noninvasive and nonionizing ultrasound-based modality that maps the electromechanical activation in all cardiac chambers at a very high frame rate (16,17). Moreover, EWI has been shown capable of accurately determining the origin of activation during ventricular pacing from different endocardial and epicardial sites in paced canine hearts in vivo (18).
In this study, EWI is used for the first time in a pediatric population. Our aim was to investigate the feasibility of using this transthoracic ultrasound technique for the localization of AP in pediatric patients with WPW.
Patient recruitment and study design
Patients presenting to the Columbia University Medical Center pediatric cardiac electrophysiology laboratory for treatment of ventricular pre-excitation by catheter ablation were approached for recruitment in this study. The Columbia University Institutional Review Board approved all methods and procedures prior to the onset of the study. After consent, background data were obtained through patient histories and review of the medical record. All patients recruited were known to have previously recorded, evident ventricular pre-excitation on resting ECG, and previously acquired transthoracic echocardiography demonstrated normal cardiac anatomy and function. Two pediatric electrophysiologists predicted the location of the AP based on the previously recorded ECG using their clinical experience and both the Boersma et al. (6) and Arruda et al. (5) algorithms.
All patients underwent EWI with a trained sonographer using standard transthoracic echocardiography immediately prior to the catheter ablation procedure. Full view of the ventricular myocardium was required for EWI processing; when the anatomy was not completely visible during the initial scan, the patient was excluded from the analysis. Obtaining an EWI scan required approximately 15 min in the pre-operative area on the day of the procedure. The processing of each EWI scan required approximately 90 min, including generation of both 2-dimensional (2D)– and 3D-rendered isochrones (approximately 70 min for 2D only). After the start of the study, the protocol was amended to include an additional EWI scan immediately after the catheter ablation procedure. After generation of the EWI isochrones, a location was assigned based on a standardized segmented template of the ventricles. This template was generated prior to the enrollment of patients and was specifically designed for this study based on similar templates in the published reports with the addition of right ventricular segments (19). This template includes 19 different segments (the basal segmentation is similar to standard ECG algorithms with 10 segments at the level of the atrioventricular rings in this study, compared with 8 segments in the Boersma et al.  algorithm and 10 in the Arruda et al.  algorithm) as shown in Figure 1E.
Both EWI and the clinical electrophysiologists reading the ECG were blinded to pre-procedure planning and the outcome of the electrophysiology study and ablation. The predicted AP locations based on the isochrones and on ECG were then compared with the site of successful ablation or the earliest site of activation if no ablation was attempted (Figures 1E to 1G). When computing the localization accuracy, predictions for both EWI and clinician interpretation of ECG were considered correct if they were in the same segment, or an adjacent segment, to the actual location of the AP.
Electromechanical wave imaging
EWI is based on a high frame rate echocardiography sequence that transmits a single diverging wave at 2,000 frames/s, while recording a lead II ECG in synchrony with the ultrasound acquisition (20). The full methods pipeline is detailed in Figures 1A to 1D. Four transthoracic apical 2D views were acquired (Figure 1A) with a 2.5-MHz phased array transducer (P4-2; ATL/Philips, Andover, Massachusetts) connected to a Vantage Research scanner system (Verasonics Inc., Kirkland, Washington). A 90° and 14-cm deep field of view was used to image the ventricles. However, for adolescents older than 16 years old it was necessary to perform the scans with a larger 20-cm depth, standardly used in adults, in order to cover the entire region of interest.
Manual segmentation of the myocardium was performed on the first B-mode frame for each view and tracked automatically throughout the rest of the cardiac cycle (Figure 1B) (21). Motion estimation was performed axially on the radiofrequency data with 1D cross-correlation tracking (22). The incremental axial strains were then derived with a least-squares estimator and overlaid onto the B-mode images (23). The ventricular activation times were defined as the zero-crossing of the strain curves, that is, the timing of the first sign change in interframe electromechanical axial strain after the QRS onset (24). The zero-crossing locations were picked for ∼100 randomly selected points in the segmented myocardial region of interest (Figure 1C), and the activation times were then interpolated throughout the entire mask to achieve a homogeneous pattern. All 2D isochrones display the electromechanical activation in milliseconds, with the earliest activated region in red and the latest in blue. The 4 resulting multi-2D electromechanical activation maps were later coregistered around the left ventricle longitudinal axis of symmetry. In each longitudinal slice, a linear interpolation was performed around the circumference to subsequently generate the 3D-rendered isochrones (Figure 1D) (25).
Electrophysiology study and ablation procedures
All catheter ablation procedures were performed under general anesthesia using standard techniques, equipment, and electroanatomic mapping (EnSite, St. Jude Medical, Inc., St. Paul, Minnesota). All patients had a surface ECG recorded followed by vascular access. After vascular access was obtained, catheters were placed near the His bundle, right atrial appendage, right ventricular apex, and coronary sinus. Pacing protocols were performed with rapid atrial pacing from the high right atrium, atrial and ventricular extrastimulus testing at baseline, and on isoproterenol. For left-sided AP, access was obtained via transseptal puncture. Ablation was performed with radiofrequency or cryoablation technique. AP location was determined by the site of successful ablation.
Data were reported as a frequency (percentage), median (interquartile range [IQR]), or mean ± SD as appropriate. Comparisons of EWI and ECG predictions to electrophysiology study and ablation results are shown on correlation maps. Heat maps of the correlation tables were generated using GraphPad Prism version 7.03 for Windows, (GraphPad Software, La Jolla, California). EWI and ECG localization performances were also quantified with general accuracy and segment-specific positive predictive value and sensitivity analysis.
Between March 21, 2017 and May 29, 2018, 15 pediatric patients with ventricular pre-excitation on 12-lead ECG were consented for the study. All patients presented for ablation of ventricular pre-excitation at the Columbia University Medical Center pediatric electrophysiology laboratory. All 15 patients underwent transthoracic imaging with a trained sonographer. One patient was excluded for the inability to image the entire ventricular myocardium due to a poor acoustic window. The mean age of the cohort was 13.8 ± 2.8 years and 50% were male. Baseline characteristics of the patients are shown in Table 1. Six patients also underwent EWI scans after their catheter ablation procedures.
Catheter mapping and ablation demonstrated a single AP in all 14 included patients. Specific locations based on our template (as seen in Figure 1E) included 3 left lateral, 2 left posterolateral, 5 posteroseptal, 1 anteroseptal, 1 right posterolateral, 1 right anterior, and 1 fasciculoventricular pathways with the earliest ventricular activation in the mid-septal right ventricle. The identified fasciculoventricular pathway was not ablated. Of the 13 patients for whom ablation was attempted, all 13 AP (100%) were successfully ablated. The patient with the anteroseptal pathway was initially not ablated secondary to mechanical disruption of the pathway preventing accurate intracardiac mapping. He subsequently returned to the laboratory and was successfully ablated on the second attempt. Ablation was performed with radiofrequency current in 12 patients and with cryoablation technique in 1 patient. Transseptal puncture was performed on 6 patients. Median fluoroscopy time was 0.3 (IQR: 0.1 to 3.5) min and dose of radiation was 13.2 (IQR: 7.0 to 70.0) μGym2 and 1 patient underwent ablation without fluoroscopy. The means and ranges of PR, QRS, AH, and HV intervals as determined by baseline measurements in the electrophysiology laboratory are also described in Table 1.
EWI and ECG predictions
EWI predicted 14 of the 14 AP locations (100%) by correctly localizing the areas of earliest ventricular activation using the segments described in the methods section. Examples of EWI isochrones are shown in the following figures: left lateral AP before and after ablation in Figure 2; right posterolateral AP before and after ablation in Figure 3; a posteroseptal AP before ablation in Figure 4; and a fasciculoventricular AP in Figure 5.
ECG analysis correctly predicted 11 of the 14 AP locations (78.6%) using both the Boersma et al. (6) and Arruda et al. (5) algorithms, respectively (predictions in immediately adjacent segments were considered correct). It should be noted that most algorithmic analyses, including these, do not include segments distal to the atrioventricular rings and therefore are not applicable for prediction of the fasciculoventricular pathway (5,6).
Correlation heat maps of AP location prediction with EWI versus ECG are shown in Figure 6. More quantitatively, positive predictive value and sensitivity analysis is provided in Table 2 for each ventricular segment and quantifies the AP localization performances of EWI versus both the Boersma et al. (6) and Arruda et al. (5) ECG algorithms.
There were no complications during EWI scans and no major complications during catheter ablations.
In this cohort, EWI was capable of both localizing and visualizing the earliest ventricular activation in 14 of the 14 included patients prior to the catheter ablation procedures. The patients were selected because they were presenting for catheter ablation. EWI localization was more accurate than ECG analysis with 2 different algorithms in our cohort.
The safety and efficacy of WPW ablation is well documented, but approximately 6% of ablations are still unsuccessful. This is variable by pathway location, from a 98% success rate for left free wall pathways to 88% to 89% for septal pathways (3). Complications from catheter ablation of AP are rare but can still occur. In addition, the risks of WPW ablation can vary by location, such as atrioventricular block in septal pathways, complications from transseptal puncture in left-sided ablation, and differences in fluoroscopy and anesthesia times based on pathway location. Some less common pathways, such as the fasciculoventricular AP, do not require catheter ablation at all. Having knowledge of the location of the pathway is crucial for both planning of catheter ablation and patient counseling prior to the procedure. Adding EWI to the standard 12-lead ECG has the potential to increase the accuracy of AP localization prior to catheter ablation procedures.
In this cohort, EWI was capable of localizing AP in a variety of locations to an approximately 1- to 2-cm area of myocardium in each case (see scale bars on Figures 2, 3, 4, and 5). EWI succeeded with a high number of anatomic segments surrounding the atrioventricular rings; this number was consistent with or greater than 12-lead ECG algorithms (10 segments at the level of the atrioventricular rings for EWI in this study, as seen in Figure 1E compared with 8 for the Boersma et al.  algorithm and 10 for the Arruda et al.  algorithm) (4,8). In addition, EWI was able to correctly identify the earliest area of ventricular pre-excitation from a fasciculoventricular pathway. The latter was localized distal to the atrioventricular ring in the mid-ventricular septum, and no current AP localization algorithm would predict this location. An advantage of EWI is its ability to locate pre-excitation in ventricular myocardium below the level of the atrioventricular rings. EWI has also been shown to be capable of localizing premature ventricular contractions in a single patient and accurately illustrating the propagation of atrial activation in normal sinus rhythm (18).
The exact spatial resolution is dependent on both the quality of imaging and the location of the pathway and is illustrated on a case-by-case basis with scale bars in each figure. When using 2D echocardiography, the location of the myocardial points imaged can affect the specificity of the EWI results. For example, as seen in the methods (Figure 1D), there are 8 image samples around the left ventricle but only 4 in the right ventricle. This inherently means that the spatial resolution of EWI will be higher for left-sided AP. Nevertheless, EWI was capable of localizing all pathways regardless of location, assuming good quality echocardiography.
The degree of pre-excitation did not affect the accuracy of EWI in this cohort. Whereas most patients were substantially pre-excited (as described in Table 1), the presence of less obvious pre-excitation did not affect the resulting localization. For example, 1 patient with a left posterolateral pathway (baseline intervals: PR: 118 ms; QRS: 103 ms; AH: 46 ms; HV: 28 ms) was successfully imaged and localized with EWI, suggesting the usefulness of EWI in patients with minimal pre-excitation.
Limitations of EWI
EWI is primarily limited by its reliance on high-quality ultrasound imaging. The myocardium is required to be fully visible in the views before EWI processing can be applied. The 1 excluded patient in which EWI could not be performed was a female adolescent, where breast tissue resulted in a difficult acoustic window and the entire ventricular myocardium could not be imaged. Even with high-quality echocardiography, certain anatomical areas are more difficult to image in multiple views. For example, the right ventricle has half the image sampling as the left ventricle, as seen in the methods section (Figure 1D). Because EWI relies on having the area of interest within view, this might limit the spatial resolution or localization altogether. This undersampling phenomenon of 2D echocardiography could potentially be overcome by using EWI with true 3D ultrasound, which would allow for imaging of all myocardium within the field of view. This has currently been shown to be feasible in open-chest canine studies and healthy volunteers (26). EWI is dependent on the presence of anterograde ventricular pre-excitation to detect the location of the AP. As demonstrated in this study, even minimal pre-excitation is sufficient, but this technique is obviously not applicable in patients with concealed AP.
This is a study in a small cohort of 15 WPW patients imaged with EWI. This study demonstrates the technique and suggests its potential uses, but the limited sample size prevents analysis on clinical measures, and this study does not comment on the effect of EWI on clinical outcomes. The blinding of the treating electrophysiologist to the EWI results prevents analysis of EWI’s effect on the procedures. Given the limited number of patients, certain pathway locations were not included. In addition, this is a single-center study of pediatric patients. This was a select population deemed suitable and selected for catheter ablation; therefore these results may not be generalizable to all patients with ventricular pre-excitation. Given the small sample size, further study is needed to both validate EWI and determine its effect in clinical practice.
EWI was shown to be capable of consistently localizing AP in variable locations more frequently than the 12-lead ECG did in a pediatric patient population. A higher correlation was obtained between the electroanatomic mapping results and EWI predictions than against ECG predictions. EWI isochrones can also provide more detailed anatomical visualization. These findings indicate that this modality has the potential to better inform a treating electrophysiologist on the precise AP location in pre-procedural planning as well as post-procedural assessment.
COMPETENCY IN MEDICAL KNOWLEDGE: This work has implications and applications in the areas of patient care and procedure skills.
TRANSLATIONAL OUTLOOK: This study is the first to use EWI for localization of AP in pediatric patients. Previously published EWI manuscripts have a few selected adult cases. This study has a larger cohort than previous studies and is the first pediatric study. However, more validation is required for EWI in AP, and EWI’s potential use in arrhythmias needs further investigation.
The authors express their very great appreciation to Vincent Sayseng, MS, and Koki Nakanishi, MD, for their time and valuable assistance acquiring part of the data. The authors also thank Julien Grondin, PhD, for his helpful discussions.
↵∗ Mrs. Melki and Mr. Grubb contributed equally to this work and are joint first authors.
↵† Drs. Liberman and Konofagou contributed equally to this work and are joint senior authors.
Supported in part by the National Institutes of Health grant nos. R01 HL114358, R01 HL140646-01, and R01 EB006042. All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All 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
- accessory pathway
- electromechanical wave imaging
- interquartile range
- Received August 15, 2018.
- Revision received December 2, 2018.
- Accepted December 4, 2018.
- 2019 American College of Cardiology Foundation
- Pappone C.,
- Vicedomini G.,
- Manguso F.,
- et al.
- Saul J.P.,
- Kanter R.J.,
- Abrams D.,
- et al.
- Wren C.,
- Vogel M.,
- Lord S.,
- et al.
- Botvinick E.,
- Frais M.,
- O'Connell W.,
- et al.
- Ghosh S.,
- Rhee E.K.,
- Avari J.N.,
- Woodard P.K.,
- Rudy Y.
- Ishizu T.,
- Seo Y.,
- Igarashi M.,
- et al.
- Esmaeilzadeh M.,
- Omran M.T.,
- Maleki M.,
- et al.
- Provost J.,
- Ghambir A.,
- Vest J.,
- Garan H.,
- Konofagou E.E.
- Costet A.,
- Wan E.,
- Bunting A.,
- Grondin J.,
- Garan H.,
- Konofagou E.
- Costet A.,
- Wan E.,
- Melki L.,
- et al.
- Cerqueira M.D.,
- Weissman N.J.,
- Dilsizian V.,
- et al.,
- for the American Heart Association Writing Group on Myocardial Segmentation and Registration for Cardiac Imaging
- Grondin J.,
- Costet A.,
- Bunting E.,
- et al.
- Melki L.,
- Costet A.,
- Konofagou E.E.
- Nauleau P.,
- Melki L.,
- Wan E.,
- Konofagou E.
- ↵Grondin J, Wang D, Trayanova N, Konofagou E. 3-D electromechanical activation mapping of the heart in canines and humans in vivo. Paper presented at: 2017 International Congress on Ultrasonics; December 18-20, 2017; Honolulu, Hawaii.Grondin J, Wang D, Trayanova N, Konofagou E. 3-D electromechanical activation mapping of the heart in canines and humans in vivo. Paper presented at: 2017 International Congress on Ultrasonics; December 18-20, 2017; Honolulu, Hawaii.