- Email: [email protected]
Activation patterns during selective pacing of the left ventricle can be characterized using noninvasive electrocardiographic imaging
Available online at www.sciencedirect.com
Journal of Electrocardiology 40 (2007) S111 – S117 www.jecgonline.com
Activation patterns during selective pacing of the left ventricle can be characterized using noninvasive electrocardiographic imaging Heather Joanne Shannon, MRCP, a,b,⁎ Cesar O. Navarro, PhD, a Bernie A. Smith, RCN, a Anthony J. McClelland, MD, a Ernest W. Lau, MD, a Michael J.D. Roberts, MD, a John M.C. Anderson, DPhil, c Jeninifer A. Adgey, MD a a
Regional Medical Cardiology Centre, Royal Victoria Hospital, Belfast, Northern Ireland, United Kingdom b Queens University of Belfast, Northern Ireland, United Kingdom c University of Ulster at Jordanstown, Belfast, Northern Ireland, United Kingdom Received 13 April 2007; revised 21 May 2007; accepted 30 May 2007
Abstract
Background: Noncontact endocardial mapping allows accurate beat-to-beat reconstruction of the reentrant pathway of ventricular tachycardia and improves outcomes after ablation. Several studies support electrocardiographic imaging (ECGI) as a means of noninvasively outlining epicardial activation despite constraints of internal geometry. However, few have explored its clinical application. This study aims to evaluate ECGI during selective left ventricular (LV) pacing, relative to an invasive approach. Methods: Multisite pacing was performed within the left ventricles of 3 patients undergoing invasive procedures. Simultaneous recording of endocardial potentials using a noncontact multielectrode array and body surface potentials (BSP) using an 80-electrode torso vest was performed. A total of 16 recordings were made. The inverse solution was applied to BSP to reconstruct epicardial activation. Single-paced beats from real and virtual electrograms were used to construct 3dimensional isochronal and isopotential maps. Endocardial and epicardial data were then superimposed onto a single geometry to allow quantitative comparison of activation foci. Results: Good correlation was observed between endocardial activation patterns and those reconstructed from BSP using ECGI. This was repeatedly demonstrated in all LV regions except for the septum (3 recordings). Epicardial isochronal maps were able to locate early and late activation to mean distances of 13.8 ± 4.7 and 12.5 ± 3.7 mm from endocardial data. Isopotential maps localized pacing sites with comparable accuracy (14 ± 5.3 mm). Conclusions: Body surface potentials and reconstructed epicardial activation patterns during LV pacing correlate well with endocardial data acquired invasively. The exception was during pacing of the septum. Although early results are encouraging, further quantitative data are required to fully validate and apply this noninvasive tool in the clinical arena. © 2007 Elsevier Inc. All rights reserved.
Keywords:
Electrocardiographic imaging; Left ventricular pacing; Noncontact endocardial mapping
Introduction The pattern of spread of electrical activation within the left ventricle (LV) is of physiological and clinical interest. In a patient with a damaged heart (such as by an ischemic or other insult), the pattern of electrical activation during sinus rhythm ⁎ Corresponding author. Regional Medical Cardiology Centre, Royal Victoria Hospital, Belfast, BT12 6BA Northern Ireland, United Kingdom. Tel.: +44 28 90632171; fax: +44 28 90312907. E-mail address: [email protected] 0022-0736/$ – see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jelectrocard.2007.05.034
may reveal electrical dysynchrony within the LV and help identify patients who may benefit from resynchronization therapy. Alternatively, during ventricular tachycardia (VT), which remains one of the most challenging of arrhythmias to treat definitively, the pattern of electrical activation will aid diagnosis of the tachycardia mechanism and facilitate localization and ablation of the arrhythmia substrate. Noncontact endocardial mapping was first described in 1987 by Taccardi et al1 and has been the subject of extensive research over the last decade. The technique has expanded
S112
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
the indications for VT ablation2 and improved clinical outcomes through its ability to capture global LV endocardial activation patterns from a single ventricular beat. However, the requirement of at least 1 catheter in the LV cavity means it can only be accomplished invasively. Electrocardiographic imaging (ECGI) is a noninvasive means of detailing the spatiotemporal pattern of cardiac electrical activity using measurements derived from the body surface. Like noncontact mapping, the global cardiac electrical activation pattern can be captured from just a single beat. Unlike noncontact mapping, the electrical signal recorded at the body surface is not of either an endocardial or an epicardial potential but that of global transmural activation. Several studies have shown that ECGI can reconstruct epicardial potentials, electrograms, and isochrones over the entire epicardial surface during the cardiac cycle, providing detailed information on local activation.3-5 One potential clinical use is in the beat-to-beat mapping of dynamically changing arrhythmias such as polymorphic VT.4 Epicardial data acquired noninvasively would undoubtedly be useful for catheter localization and ablation of VT, where the existence of multiple foci, at variable intramural depths, are widely recognized phenomena. The LV myocardium is a 3-dimensional (3-D) structure with a considerable transmural muscle thickness (≈1 cm in the normal adult heart). The endocardial surface area is smaller than the epicardial surface and hence there cannot be a precise 1:1 relationship between points on the 2 surfaces. Consequently, LV endocardial and epicardial electrical activations are related but separate events. Over the last decade, work in the field of invasive electrophysiological mapping has focused on validating the computerized noncontact mapping technique using a multielectrode array (MEA), with very encouraging results.2,6-10 Similarly with ECGI, investigators have concentrated on improving inverse mathematical techniques and validating reconstructed epicardial data relative to that measured directly.3-5,11-13 Most of the work have involved animal models. The aim of the current study is to extend the use of ECGI to the clinical arena by conducting a qualitative comparison of reconstructed epicardial and endocardial data. Comparison of stimulation sites and activation patterns at epicardial and endocardial level is performed by simultaneously mapping using invasive and noninvasive modalities during selective pacing of the LV.
right coronary artery lesion. The second patient had normal coronary arteries and a structurally normal heart but required pulmonary vein isolation for symptomatic paroxysmal atrial fibrillation. The third patient required diagnostic coronary angiography, which showed triple-vessel coronary disease unsuitable for percutaneous or surgical intervention. In all cases, left ventricular systolic function was within normal limits (ejection fraction N50%) on echocardiography. The study was approved by the Research Ethics Committee of Northern Ireland. All patients were studied in the postabsorptive state after giving written informed consent. Mapping procedures Noncontact mapping system The EnSite 3000 (Endocardial Solutions, Inc, St. Paul, MN) computerized mapping system was used. A 9F 64electrode noncontact balloon catheter (MEA) was positioned in the LV using a retrograde aortic approach as previously described.2,6,7 A conventional 7F quadripolar electrode catheter for mapping and ablation was placed in the LV using a transeptal approach. All patients were anticoagulated using intravenous heparin before deployment of the array, with additional boluses as required to maintain the activated clotting time at 300 to 400 seconds. The MEA was used to construct the 3-D geometry of the LV endocardium (virtual endocardium) by providing a locator system to identify the position of the conventional roving catheter and computing 3360 virtual endocardial unipolar electrograms. These were calculated by customized software based on an inverse solution to Laplace's equation by use of a boundary element method.6 Knowledge of the electrical potential at each virtual electrode at any given moment in time enabled construction of 3-D isopotential maps. The use of color to observe successive maps throughout the cardiac cycle enabled observation of the depolarization wavefront. The timing at the point of the maximum −dV/dt of the QRS was taken as a marker of intrinsic activation (breakthrough).6 In addition, the activation time for each virtual electrode allowed construction of isochronal maps outlining progression of the activation wavefront over time. These were used to determine regions of earliest and latest activation with crowding of isochrones indicating conduction block. Electrocardiographic imaging
Methods Patients Three consecutive patients undergoing invasive cardiological procedures were studied. All were male with a mean age of 61.3 years (range 54-69 years). The first patient required an automated implantable defibrillator replacement, with the initial implant 3 years previously following a ventricular fibrillation cardiac arrest in the setting of an acute inferior myocardial infarction. Coronary angiography at the time showed single-vessel disease, and he subsequently underwent percutaneous intervention to a culprit
Body surface potentials (BSPs) were recorded using PRIME™ ECG, which has previously been described.5 The system is composed of a flexible plastic, 80-electrode harness applied to the anterior and posterior torso along with a portable recording unit. A total of 64 electrodes are positioned anteriorly (including 3 proximal limb leads) and 16 posteriorly, enabling recording of 80 simultaneous unipolar ECG signals with respect to the Wilson Central Terminal. Standard 12-lead electrodes form a subset of the 80-lead configuration. Geometric information was obtained using images of slices sectioned perpendicularly to the axial axis of
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
S113
Table 1 Comparison of activation timings and calculated distances between the first electrode to be activated at endocardial and epicardial level (16 paced recordings) Pacing region
Anterolateral Apex Apex Apex Apex Apex Posterior Posterior Posterior Posterior Posterolateral Posterolateral Posterolateral Septum Septum Septum
Total activation time (ms)
Time to earliest activation (ms)
Calculated interelectrode distance between earliest activated electrodes (mm)
Endocardium
Epicardium
Endocardium
Epicardium
Isochronal maps
Isopotential maps
55 41 107 107 96 75 70 53 60 90 88 85 80 54 89 70
67 118 136 127 143 135 139 146 116 95 106 104 187 88 110 83
5 0 3 8 14 0 10 7 0 0 2 0 0 6 1 0
25 0 9 8 25 0 10 7 0 20 25 15 15 35 20 48
15.5 13.5 14.2 15.5 6.2 10.9 10.8 16.0 9.2 19.6 14.1 16.3 9.7 10.2 13.6 26.0
6.0 7.5 21.8 16.4 10.7 9.2 22.3 19.7 17.6 17.4 5.8 15.5 12.1 11.2 16.3 14.0
Calculated distance between latest activated electrodes on isochronal maps (mm) 18.6 7.1 13.1 11.2 11.8 5.7 13.5 13.1 12.7 14.3 9.0 13.7 10.6 9.4 16.1 19.5
The total activation time is defined as the time difference between the earliest and latest electrodes to be activated. The time to earliest activation is defined as the time taken for the earliest maximum −dV/dt to occur after the pacing stimulus. The calculated distance between the earliest activated electrodes refers to the distance between the first electrode to be activated on the endocardium and the first electrode to be activated on the epicardium. This was calculated for both isochronal and isopotential maps. The calculated distance between the latest activated electrodes refers to the distance between the last electrode to be activated on the endocardium and the epicardium. This was calculated for isochronal maps.
a male cadaver from The Visible Human Project.14 The contours of the heart, lungs, sternum, spine, subcutaneous tissue, and torso surface were obtained by means of manual segmentation. The segmented slices were then stacked to create a 3-D representation of the torso, and Computer Aided Design software was used to construct a 3-D mesh.15 The inverse solution based on the finite element method was applied to BSPs to reconstruct 59 potentials at epicardial level.16 Isopotential and isochronal maps could then be displayed on the epicardial geometry.5,17 As with reconstructed endocardial data, the site of earliest activation corresponded to the region of most negative potential after the pacing spike5 or to the point when the earliest maximum −dV/dt of the QRS occurred.4 Pacing protocol Pacing was performed using the conventional mapping catheter by locating it to multiple anatomical sites on the left ventricular endocardium. A cycle length of 600 milliseconds with pacing output of 5 V ms−1 was used. Noncontact MEA and body surface data were recorded simultaneously by 2 investigators with both recordings initiated simultaneously. Reconstructed endocardial electrograms and those from the body surface were later reviewed, and by coordinating exact timings in milliseconds and QRS morphology in a corresponding chest lead, the same beat from each system was selected for analysis.
Data analysis A total of 16 paced recordings were made. Once the corresponding paced beat was selected from endocardial and epicardial electrograms, the maximum −dV/dt of the QRS from each was used to display dynamic 3-D isopotential maps. The maps were similarly color-coded from ‘red,’ as an indicator of the minimum amplitude, to ‘blue,’ as the maximum. This enabled distinction of the region of breakthrough, which was taken to correlate with the site of pacing.3,5 In the same way, isochronal maps were displayed on the 3-D geometries, with color progression from red to blue reflecting activation with time. Regions of crowded isochrones indicative of scar tissue could subsequently be compared at both levels. Data were thus qualitative. It was therefore decided to derive a semiquantitative comparison of pacing sites at endocardial and epicardial levels. Each tailored endocardial LV geometry was integrated into the corresponding epicardial geometry in the torso by aligning the endocardia using their centroids. The electrode showing the earliest negative deflection (activation) was determined for each. A straight line was constructed from the centroid of the endocardial geometry to the appropriate electrode on the epicardial surface. The intersection with the endocardium was marked, and the distance between this point and the earliest activated endocardial electrode was calculated directly. This was used to provide an estimate of the accuracy of the noninvasive modality.
S114
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
Fig. 1. Comparison of isochronal (A, B, and C) and isopotential (D, E, and F) maps at endocardial, body surface, and epicardial levels during pacing of the LV apex (red, earliest activation). The resolution can be seen to be higher using isopotentials. The correlation between all 3 maps is good during pacing of this region. The color legend (time in milliseconds for isochronals and amplitudes in millivolts for isopotential maps) are displayed on the left. Amplitude scales for body surface and epicardial isopotential maps have been normalized to enable comparison of early activation with endocardial maps. All views are anteroposterior. The position of the right ventricle is indicated by ‘RV'.
Statistical analysis The mean and SD were used for activation times and distances between pacing sites at both endocardial and epicardial levels.
Results Sixteen paced recordings were made—5 from patient 1, 7 from patient 2, and 4 from patient 3. Left ventricular pacing sites were divided anatomically into 5 regions and marked on the endocardial geometry: posterior (4 cases), posterolateral (3 cases), anterolateral (1 case), apical (5 cases), and septal (3 cases) (Table 1). Good correlation was observed between regions of breakthrough on the endocardium and epicardium, with higher resolution seen on isopotential maps. These regions coincided with pacing sites as marked on the endocardial geometry by the pacing catheter. Fig. 1 shows a comparison of reconstructed endocardial maps acquired
invasively and body surface with reconstructed epicardial maps acquired noninvasively during pacing of the LV apex. The correlation, qualitatively, can be seen to be good. This finding was repeatedly demonstrated in all regions of the LV except for the septum. Of the 3 recordings made in this region, all were located using the MEA. By contrast, reconstructed epicardial isopotential and isochronal maps located the pacing stimulus to the posterior anatomical segment (Fig. 2). Total activation timings were greater using reconstructed epicardial compared with endocardial isochrones (mean of 118.8 milliseconds and SD of 29.4 vs mean of 76.3 milliseconds and SD of 19.9) (Table 1). Similarly, the mean time from application of the pacing stimulus to initial activation was greater at epicardial than at endocardial level (16.4 vs 3.5 milliseconds) (Table 1). The endocardial-epicardial delay (time difference between the onset of endocardial and epicardial activation) was greater when the stimulus was applied to the septum (mean delay, 32.0 milliseconds) compared with other regions (mean delay, 8.5 milliseconds).
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
S115
Fig. 2. Comparison of isochronal (A, B, and C) and isopotential maps (D, E, and F) at endocardial, body surface, and epicardial levels during pacing of the LV septum (red, earliest activation). The resolution can be seen to be higher using isopotentials. Although septal activation can be clearly seen on the endocardial maps, body surface and reconstructed epicardial maps are poor at locating pacing in this region. The color legend (time in milliseconds for isochronals and amplitudes in millivolts for isopotential maps) are displayed on the left. Amplitude scales for body surface and epicardial isopotential maps have been normalized to enable comparison with endocardial maps. All views are anteroposterior. The position of the right ventricle is indicated by ‘RV'.
In the absence of pacing, 2 additional baseline map recordings were made during sinus rhythm in patients 1 and 3, and activation was seen to propagate from the endocardium to the epicardium. Electrocardiographic imaging was as effective at outlining the direction of activation (clockwise vs anticlockwise) as reconstructed endocardial data. However, the spread of activation differed at both levels when comparing isopotential and isochronal maps. Qualitatively, correlation was observed between regions of crowded isochrones at endocardial and epicardial level with 2 exceptions: when the pacing stimulus was at the septum and when regions of slow conduction were seen endocardially in the region of the septum. Reconstructed epicardial data were poor at determining activity in this region. Further quantitative analysis with integration of endocardial into reconstructed epicardial geometry was then performed. Table 1 shows the calculated distances between the first electrode to be activated using both systems. Comparative data for isochronal and isopotential maps are shown for all pacing sites. Epicardial isochronal maps were able to locate stimulation sites to a mean of 13.8 mm (SD, 4.7)
from corresponding endocardial data. Isopotential maps showed similar accuracy (mean, 14.0 mm; SD, 5.3). Despite the poor qualitative correlation during septal pacing, the interelectrode distances in this region were not significantly greater than during pacing of other regions (mean, 16.6 mm by isochrones and 13.8 mm by isopotentials). Epicardial isochronal maps were also able to locate regions of latest activation to a mean of 12.5 mm (SD, 3.7) from endocardial data (Table 1). Discussion This is the first clinical pacing study to compare epicardial data acquired noninvasively using ECGI with endocardial data acquired invasively using a noncontact mapping system. The data suggest that ECGI is as effective at locating LV endocardial pacing as invasive data except when the stimulus is located at a septal origin. Oster et al4 have shown that, by epicardial pacing of a dog heart in a human torso–shaped tank, single pacing sites over the epicardial surface could be reconstructed to within 10 mm or less of their measured
S116
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
positions. They also showed that regions of sparse and crowded isochrones could be accurately reconstructed using ECGI. Although problems with septal pacing have not previously been documented, there have been few data recorded in the clinical area. In addition, most studies have used epicardial pacing, making definitively septal stimulation difficult. Oster et al4 looked at anterior and posterolateral sites predominantly, whereas in a more recent study by McClelland et al,5 anterolateral, posterolateral, and posterior pacing sites were examined. Further work with pacing in all 5 segments documented in this study, particularly the septum, is undoubtedly required to substantiate these preliminary results. Further data are required to enable explanation of the discrepancy at the septum. One possibility relates to the structure and orientation of endocardium within the epicardial model. During pacing of the lateral wall, for example, the activation wavefront has only a short depth of myocardium to traverse before becoming apparent at epicardial level. Conversely, a pacing stimulus applied at the endocardial septum is further from all epicardial surfaces, let alone a well-defined septal wall on the 3-D geometry. It seems the activation wavefront could pass in 1 of 3 directions: (1) posteriorly to become apparent at the posterior epicardial wall, (2) towards the anterior wall, or (3) towards the right ventricle. In all 3 of our cases, the site of earliest activation was seen at the posterior epicardial wall. Interestingly, the endocardial wavefront in each case reached the posterior wall after leaving the septum, and there seemed to be a greater delay in initial epicardial activation (mean, 32.0 milliseconds). This compares to a mean delay of 8.5 milliseconds during pacing at alternative sites. Therefore, it might be the case that ECGI is unable to identify any activity at the septum, or alternatively, that septal activity is reconstructed with some delay at the nearest epicardial surface of the 3-D model. The attempt to quantitatively assess pacing site correlation has clear limitations. Anatomical segmentation into 5 regions was performed, not by surface area but according to regions defined by echocardiography. Therefore, the septum and apex, for example, represent a much smaller area than the posterior segment. This may explain why qualitatively, during pacing of the septum, the correlation seems poor, yet absolute interelectrode distances are not significantly greater than in other regions. By analyzing maps recorded in the absence of pacing, we were able to support the well-documented phenomenon that, in sinus rhythm, propagation of activation is mostly transmural, from endocardium to epicardium.3 Oster et al3 have previously shown that during pacing, the direction of the propagated activation wavefront varies according to stimulation depth and is dependent on myocardial fiber orientation. Again using an isolated dog heart suspended in a human torso tank, they demonstrated a clockwise rotation of the wavefront with subendocardial pacing and counterclockwise rotation with epicardial stimulation. From our study, performed in real time with human subjects and using endocardial/subendocardial pacing alone, we were unable to show clockwise rotation
exclusively for the activation wavefront. Only 6 of the 16 recordings supported these findings, with 2 cases showing predominantly counterclockwise rotation and the remaining 8 showing multidirectional propagation. We propose an important potential clinical application of ECGI is the anatomical localization of VT to aid ablation. Current long-term outcomes for VT ablation remain suboptimal (partly due to localization, but other variables including multiple sites and incomplete lesions transmurally) despite the advent of noncontact endocardial mapping, which has enabled even poorly tolerated arrhythmias to be treated.2 This is primarily due to the complexity of the VT circuit, a high proportion of which involves 3-D reentry18-20 or focal epicardial origin. Although the exit site has traditionally been targeted for ablation, recent work has shown the importance of locating the diastolic component of the circuit. Using the noncontact MEA in a clinical setting, Schilling et al7 were able to identify a complete tachycardia circuit within the diastolic interval in only 21%. In another study, this was identified in only 1 VT.2 One explanation is that at least part of the diastolic component is intramural or epicardial and therefore cannot be identified by the system.7 Burnes et al13 were the first to demonstrate the noninvasive reconstruction of reentrant activation at epicardial level during VT in a canine model. Combined acquisition of invasive endocardial data and reconstructed epicardial data would provide more comprehensive assessment of the VT diastolic circuit and exit site before radiofrequency ablation. Study limitations The main limitation of this study is the small numbers involved, with 16 recordings from only 3 patients. It is therefore with caution that we propose any definitive conclusion, rather emphasizing the need for further work. The limitations of ECGI are well documented, with torso inhomogeneities a major source of error.3 In addition, the inverse solution applied to BSPs used a general torso model rather than one tailored to individual body habitus. The ability to import patient-specific LV geometries by MRI/CT into the Prime reconstruct would be the obvious next step. Despite geometrical limitations, some studies have indicated that torso inhomogeneities affect only epicardial potential magnitudes and not isochronal patterns of activation.3,12 Finally, the use of reconstructed rather than directly measured data at epicardial and endocardial levels creates a further source of error. Conclusions This is the first study to support the use of ECGI in determining the site of endocardial pacing when compared directly with noncontact endocardial data. The small numbers employed in this study predict ECGI to be comparable to invasive endocardial data in all regions except during direct stimulation of the septum. Further data collection is essential before forming any definitive conclusion. We propose the combined use of ECGI and noncontact endocardial mapping to detail the complete diastolic circuit of VT, with the potential for guiding radiofrequency ablation of this difficult tachyarrhythmia.
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
References 1. Taccardi B, Arisi G, Macchi E, Baruffi S, Spaggiari S. A new intracavitary probe for detecting the site of origin of ectopic ventricular beats during one cardiac cycle. Circulation 1987;75:272. 2. Strickberger SA, Knight BP, Michaud GF, Pelosi F, Morady F. Mapping and ablation of ventricular tachycardia guided by virtual electrograms using a noncontact, computerized mapping system. J Am Coll Cardiol 2000;35:414. 3. Oster HS, Taccardi B, Lux RL, Ershler PR, Rudy Y. Electrocardiographic imaging: noninvasive characterization of intramural myocardial activation from inverse-reconstructed epicardial potentials and electrograms. Circulation 1998;97:1496. 4. Oster HS, Taccardi B, Lux RL, Ershler PR, Rudy Y. Non invasive electrocardiographic imaging: reconstruction of epicardial potentials, electrograms and isochrones and localization of single and multiple electrocardiac events. Circulation 1997;96:1012. 5. McClelland AJJ, Owens CG, Navarro C, et al. Usefulness of body surface maps to demonstrate ventricular activation patterns during left ventricular pacing and reentrant activation during ventricular tachycardia in men with coronary heart disease and left ventricular dysfunction. Am J Cardiol 2006;98:591. 6. Schilling RJ, Peters NS, Davies DW. Simultaneous endocardial mapping in the human left ventricle using a noncontact catheter. Circulation 1998;98:887. 7. Schilling RJ, Peters NS, Davies DW. Mapping and ablation of ventricular tachycardia with the aid of a noncontact mapping system. Heart 1999;81:570. 8. Gornick CC, Alder SW, Pederson B, Hauck J, Budd J, Schweitzer J. Validation of a new noncontact catheter system for electroanatomic mapping of left ventricular endocardium. Circulation 1999;99:829. 9. Khoury DS, Taccardi B, Lux RL, Ershler PR, Rudy Y. Reconstruction of endocardial potentials and activation sequences from intracavitary
10.
11.
12.
13. 14. 15.
16.
17.
18.
19.
20.
S117
probe measurements: localization of pacing sites and effects of myocardial structure. Circulation 1995;91:845. Khoury DS, Berrier KL, Badruddin SM, Zoghbi WA. Three-dimensional electrophysiological imaging of the intact canine left ventricle using a noncontact multielectrode cavitary probe: study of sinus, paced, and spontaneous premature beats. Circulation 1998;97:399. Green LS, Taccardi B, Ershler PR, Lux RL. Epicardial potential mapping: effects of conducting media on isopotential and isochrone distributions. Circulation 1991;84:2513. Rudy Y, Plonsey R. A comparison of volume conductor and source geometry effects on body surface and epicardial potentials. Circ Res 1980;46:283. Burnes JE, Taccardi B, Rudy Y. A noninvasive imaging modality for cardiac arrhythmias. Circulation 2000;102:2152. Ackerman MJ. The visible human project. Proc IEEE 1998;86:504. Navarro C, Anderson J, Allen J, Toro J, Riddell J, Adgey AAJ. Construction of a model of the torso for ECG problems based on CAD tools and images of the human body. Technol Health Care 2001;9:176. Yamashita Y, Takahashi T. Use of the finite element method to determine epicardial from body surface potentials under a realistic torso model. IEEE Trans Biomed Eng 1984;BME 31:611. Owens C, Navarro C, McClelland A, et al. Improved detection of acute myocardial infarction using a diagnostic algorithm based on calculated epicardial potentials. Int J Cardiol 2006;111:292. Pogwizd SM, Hoyt RH, Saffitz JE, Corr PB, Cox JL, Cain ME. Reentrant and focal mechanisms underlying ventricular tachycardia in the human heart. Circulation 1992;86:1872. Stevenson WG, Khan H, Sager P, et al. Identification of reentry circuit sites during catheter mapping and radiofrequency ablation of ventricular tachycardia late after myocardial infarction. Circulation 1993;88:1647. Kaltenbrunner W, Cardinal R, Dubuc M, et al. Epicardial and endocardial mapping of ventricular tachycardia in patients with myocardial infarction: is the origin of the tachycardia always subendocardially localized? Circulation 1991;84:1058.
Journal of Electrocardiology 40 (2007) S111 – S117 www.jecgonline.com
Activation patterns during selective pacing of the left ventricle can be characterized using noninvasive electrocardiographic imaging Heather Joanne Shannon, MRCP, a,b,⁎ Cesar O. Navarro, PhD, a Bernie A. Smith, RCN, a Anthony J. McClelland, MD, a Ernest W. Lau, MD, a Michael J.D. Roberts, MD, a John M.C. Anderson, DPhil, c Jeninifer A. Adgey, MD a a
Regional Medical Cardiology Centre, Royal Victoria Hospital, Belfast, Northern Ireland, United Kingdom b Queens University of Belfast, Northern Ireland, United Kingdom c University of Ulster at Jordanstown, Belfast, Northern Ireland, United Kingdom Received 13 April 2007; revised 21 May 2007; accepted 30 May 2007
Abstract
Background: Noncontact endocardial mapping allows accurate beat-to-beat reconstruction of the reentrant pathway of ventricular tachycardia and improves outcomes after ablation. Several studies support electrocardiographic imaging (ECGI) as a means of noninvasively outlining epicardial activation despite constraints of internal geometry. However, few have explored its clinical application. This study aims to evaluate ECGI during selective left ventricular (LV) pacing, relative to an invasive approach. Methods: Multisite pacing was performed within the left ventricles of 3 patients undergoing invasive procedures. Simultaneous recording of endocardial potentials using a noncontact multielectrode array and body surface potentials (BSP) using an 80-electrode torso vest was performed. A total of 16 recordings were made. The inverse solution was applied to BSP to reconstruct epicardial activation. Single-paced beats from real and virtual electrograms were used to construct 3dimensional isochronal and isopotential maps. Endocardial and epicardial data were then superimposed onto a single geometry to allow quantitative comparison of activation foci. Results: Good correlation was observed between endocardial activation patterns and those reconstructed from BSP using ECGI. This was repeatedly demonstrated in all LV regions except for the septum (3 recordings). Epicardial isochronal maps were able to locate early and late activation to mean distances of 13.8 ± 4.7 and 12.5 ± 3.7 mm from endocardial data. Isopotential maps localized pacing sites with comparable accuracy (14 ± 5.3 mm). Conclusions: Body surface potentials and reconstructed epicardial activation patterns during LV pacing correlate well with endocardial data acquired invasively. The exception was during pacing of the septum. Although early results are encouraging, further quantitative data are required to fully validate and apply this noninvasive tool in the clinical arena. © 2007 Elsevier Inc. All rights reserved.
Keywords:
Electrocardiographic imaging; Left ventricular pacing; Noncontact endocardial mapping
Introduction The pattern of spread of electrical activation within the left ventricle (LV) is of physiological and clinical interest. In a patient with a damaged heart (such as by an ischemic or other insult), the pattern of electrical activation during sinus rhythm ⁎ Corresponding author. Regional Medical Cardiology Centre, Royal Victoria Hospital, Belfast, BT12 6BA Northern Ireland, United Kingdom. Tel.: +44 28 90632171; fax: +44 28 90312907. E-mail address: [email protected] 0022-0736/$ – see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jelectrocard.2007.05.034
may reveal electrical dysynchrony within the LV and help identify patients who may benefit from resynchronization therapy. Alternatively, during ventricular tachycardia (VT), which remains one of the most challenging of arrhythmias to treat definitively, the pattern of electrical activation will aid diagnosis of the tachycardia mechanism and facilitate localization and ablation of the arrhythmia substrate. Noncontact endocardial mapping was first described in 1987 by Taccardi et al1 and has been the subject of extensive research over the last decade. The technique has expanded
S112
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
the indications for VT ablation2 and improved clinical outcomes through its ability to capture global LV endocardial activation patterns from a single ventricular beat. However, the requirement of at least 1 catheter in the LV cavity means it can only be accomplished invasively. Electrocardiographic imaging (ECGI) is a noninvasive means of detailing the spatiotemporal pattern of cardiac electrical activity using measurements derived from the body surface. Like noncontact mapping, the global cardiac electrical activation pattern can be captured from just a single beat. Unlike noncontact mapping, the electrical signal recorded at the body surface is not of either an endocardial or an epicardial potential but that of global transmural activation. Several studies have shown that ECGI can reconstruct epicardial potentials, electrograms, and isochrones over the entire epicardial surface during the cardiac cycle, providing detailed information on local activation.3-5 One potential clinical use is in the beat-to-beat mapping of dynamically changing arrhythmias such as polymorphic VT.4 Epicardial data acquired noninvasively would undoubtedly be useful for catheter localization and ablation of VT, where the existence of multiple foci, at variable intramural depths, are widely recognized phenomena. The LV myocardium is a 3-dimensional (3-D) structure with a considerable transmural muscle thickness (≈1 cm in the normal adult heart). The endocardial surface area is smaller than the epicardial surface and hence there cannot be a precise 1:1 relationship between points on the 2 surfaces. Consequently, LV endocardial and epicardial electrical activations are related but separate events. Over the last decade, work in the field of invasive electrophysiological mapping has focused on validating the computerized noncontact mapping technique using a multielectrode array (MEA), with very encouraging results.2,6-10 Similarly with ECGI, investigators have concentrated on improving inverse mathematical techniques and validating reconstructed epicardial data relative to that measured directly.3-5,11-13 Most of the work have involved animal models. The aim of the current study is to extend the use of ECGI to the clinical arena by conducting a qualitative comparison of reconstructed epicardial and endocardial data. Comparison of stimulation sites and activation patterns at epicardial and endocardial level is performed by simultaneously mapping using invasive and noninvasive modalities during selective pacing of the LV.
right coronary artery lesion. The second patient had normal coronary arteries and a structurally normal heart but required pulmonary vein isolation for symptomatic paroxysmal atrial fibrillation. The third patient required diagnostic coronary angiography, which showed triple-vessel coronary disease unsuitable for percutaneous or surgical intervention. In all cases, left ventricular systolic function was within normal limits (ejection fraction N50%) on echocardiography. The study was approved by the Research Ethics Committee of Northern Ireland. All patients were studied in the postabsorptive state after giving written informed consent. Mapping procedures Noncontact mapping system The EnSite 3000 (Endocardial Solutions, Inc, St. Paul, MN) computerized mapping system was used. A 9F 64electrode noncontact balloon catheter (MEA) was positioned in the LV using a retrograde aortic approach as previously described.2,6,7 A conventional 7F quadripolar electrode catheter for mapping and ablation was placed in the LV using a transeptal approach. All patients were anticoagulated using intravenous heparin before deployment of the array, with additional boluses as required to maintain the activated clotting time at 300 to 400 seconds. The MEA was used to construct the 3-D geometry of the LV endocardium (virtual endocardium) by providing a locator system to identify the position of the conventional roving catheter and computing 3360 virtual endocardial unipolar electrograms. These were calculated by customized software based on an inverse solution to Laplace's equation by use of a boundary element method.6 Knowledge of the electrical potential at each virtual electrode at any given moment in time enabled construction of 3-D isopotential maps. The use of color to observe successive maps throughout the cardiac cycle enabled observation of the depolarization wavefront. The timing at the point of the maximum −dV/dt of the QRS was taken as a marker of intrinsic activation (breakthrough).6 In addition, the activation time for each virtual electrode allowed construction of isochronal maps outlining progression of the activation wavefront over time. These were used to determine regions of earliest and latest activation with crowding of isochrones indicating conduction block. Electrocardiographic imaging
Methods Patients Three consecutive patients undergoing invasive cardiological procedures were studied. All were male with a mean age of 61.3 years (range 54-69 years). The first patient required an automated implantable defibrillator replacement, with the initial implant 3 years previously following a ventricular fibrillation cardiac arrest in the setting of an acute inferior myocardial infarction. Coronary angiography at the time showed single-vessel disease, and he subsequently underwent percutaneous intervention to a culprit
Body surface potentials (BSPs) were recorded using PRIME™ ECG, which has previously been described.5 The system is composed of a flexible plastic, 80-electrode harness applied to the anterior and posterior torso along with a portable recording unit. A total of 64 electrodes are positioned anteriorly (including 3 proximal limb leads) and 16 posteriorly, enabling recording of 80 simultaneous unipolar ECG signals with respect to the Wilson Central Terminal. Standard 12-lead electrodes form a subset of the 80-lead configuration. Geometric information was obtained using images of slices sectioned perpendicularly to the axial axis of
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
S113
Table 1 Comparison of activation timings and calculated distances between the first electrode to be activated at endocardial and epicardial level (16 paced recordings) Pacing region
Anterolateral Apex Apex Apex Apex Apex Posterior Posterior Posterior Posterior Posterolateral Posterolateral Posterolateral Septum Septum Septum
Total activation time (ms)
Time to earliest activation (ms)
Calculated interelectrode distance between earliest activated electrodes (mm)
Endocardium
Epicardium
Endocardium
Epicardium
Isochronal maps
Isopotential maps
55 41 107 107 96 75 70 53 60 90 88 85 80 54 89 70
67 118 136 127 143 135 139 146 116 95 106 104 187 88 110 83
5 0 3 8 14 0 10 7 0 0 2 0 0 6 1 0
25 0 9 8 25 0 10 7 0 20 25 15 15 35 20 48
15.5 13.5 14.2 15.5 6.2 10.9 10.8 16.0 9.2 19.6 14.1 16.3 9.7 10.2 13.6 26.0
6.0 7.5 21.8 16.4 10.7 9.2 22.3 19.7 17.6 17.4 5.8 15.5 12.1 11.2 16.3 14.0
Calculated distance between latest activated electrodes on isochronal maps (mm) 18.6 7.1 13.1 11.2 11.8 5.7 13.5 13.1 12.7 14.3 9.0 13.7 10.6 9.4 16.1 19.5
The total activation time is defined as the time difference between the earliest and latest electrodes to be activated. The time to earliest activation is defined as the time taken for the earliest maximum −dV/dt to occur after the pacing stimulus. The calculated distance between the earliest activated electrodes refers to the distance between the first electrode to be activated on the endocardium and the first electrode to be activated on the epicardium. This was calculated for both isochronal and isopotential maps. The calculated distance between the latest activated electrodes refers to the distance between the last electrode to be activated on the endocardium and the epicardium. This was calculated for isochronal maps.
a male cadaver from The Visible Human Project.14 The contours of the heart, lungs, sternum, spine, subcutaneous tissue, and torso surface were obtained by means of manual segmentation. The segmented slices were then stacked to create a 3-D representation of the torso, and Computer Aided Design software was used to construct a 3-D mesh.15 The inverse solution based on the finite element method was applied to BSPs to reconstruct 59 potentials at epicardial level.16 Isopotential and isochronal maps could then be displayed on the epicardial geometry.5,17 As with reconstructed endocardial data, the site of earliest activation corresponded to the region of most negative potential after the pacing spike5 or to the point when the earliest maximum −dV/dt of the QRS occurred.4 Pacing protocol Pacing was performed using the conventional mapping catheter by locating it to multiple anatomical sites on the left ventricular endocardium. A cycle length of 600 milliseconds with pacing output of 5 V ms−1 was used. Noncontact MEA and body surface data were recorded simultaneously by 2 investigators with both recordings initiated simultaneously. Reconstructed endocardial electrograms and those from the body surface were later reviewed, and by coordinating exact timings in milliseconds and QRS morphology in a corresponding chest lead, the same beat from each system was selected for analysis.
Data analysis A total of 16 paced recordings were made. Once the corresponding paced beat was selected from endocardial and epicardial electrograms, the maximum −dV/dt of the QRS from each was used to display dynamic 3-D isopotential maps. The maps were similarly color-coded from ‘red,’ as an indicator of the minimum amplitude, to ‘blue,’ as the maximum. This enabled distinction of the region of breakthrough, which was taken to correlate with the site of pacing.3,5 In the same way, isochronal maps were displayed on the 3-D geometries, with color progression from red to blue reflecting activation with time. Regions of crowded isochrones indicative of scar tissue could subsequently be compared at both levels. Data were thus qualitative. It was therefore decided to derive a semiquantitative comparison of pacing sites at endocardial and epicardial levels. Each tailored endocardial LV geometry was integrated into the corresponding epicardial geometry in the torso by aligning the endocardia using their centroids. The electrode showing the earliest negative deflection (activation) was determined for each. A straight line was constructed from the centroid of the endocardial geometry to the appropriate electrode on the epicardial surface. The intersection with the endocardium was marked, and the distance between this point and the earliest activated endocardial electrode was calculated directly. This was used to provide an estimate of the accuracy of the noninvasive modality.
S114
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
Fig. 1. Comparison of isochronal (A, B, and C) and isopotential (D, E, and F) maps at endocardial, body surface, and epicardial levels during pacing of the LV apex (red, earliest activation). The resolution can be seen to be higher using isopotentials. The correlation between all 3 maps is good during pacing of this region. The color legend (time in milliseconds for isochronals and amplitudes in millivolts for isopotential maps) are displayed on the left. Amplitude scales for body surface and epicardial isopotential maps have been normalized to enable comparison of early activation with endocardial maps. All views are anteroposterior. The position of the right ventricle is indicated by ‘RV'.
Statistical analysis The mean and SD were used for activation times and distances between pacing sites at both endocardial and epicardial levels.
Results Sixteen paced recordings were made—5 from patient 1, 7 from patient 2, and 4 from patient 3. Left ventricular pacing sites were divided anatomically into 5 regions and marked on the endocardial geometry: posterior (4 cases), posterolateral (3 cases), anterolateral (1 case), apical (5 cases), and septal (3 cases) (Table 1). Good correlation was observed between regions of breakthrough on the endocardium and epicardium, with higher resolution seen on isopotential maps. These regions coincided with pacing sites as marked on the endocardial geometry by the pacing catheter. Fig. 1 shows a comparison of reconstructed endocardial maps acquired
invasively and body surface with reconstructed epicardial maps acquired noninvasively during pacing of the LV apex. The correlation, qualitatively, can be seen to be good. This finding was repeatedly demonstrated in all regions of the LV except for the septum. Of the 3 recordings made in this region, all were located using the MEA. By contrast, reconstructed epicardial isopotential and isochronal maps located the pacing stimulus to the posterior anatomical segment (Fig. 2). Total activation timings were greater using reconstructed epicardial compared with endocardial isochrones (mean of 118.8 milliseconds and SD of 29.4 vs mean of 76.3 milliseconds and SD of 19.9) (Table 1). Similarly, the mean time from application of the pacing stimulus to initial activation was greater at epicardial than at endocardial level (16.4 vs 3.5 milliseconds) (Table 1). The endocardial-epicardial delay (time difference between the onset of endocardial and epicardial activation) was greater when the stimulus was applied to the septum (mean delay, 32.0 milliseconds) compared with other regions (mean delay, 8.5 milliseconds).
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
S115
Fig. 2. Comparison of isochronal (A, B, and C) and isopotential maps (D, E, and F) at endocardial, body surface, and epicardial levels during pacing of the LV septum (red, earliest activation). The resolution can be seen to be higher using isopotentials. Although septal activation can be clearly seen on the endocardial maps, body surface and reconstructed epicardial maps are poor at locating pacing in this region. The color legend (time in milliseconds for isochronals and amplitudes in millivolts for isopotential maps) are displayed on the left. Amplitude scales for body surface and epicardial isopotential maps have been normalized to enable comparison with endocardial maps. All views are anteroposterior. The position of the right ventricle is indicated by ‘RV'.
In the absence of pacing, 2 additional baseline map recordings were made during sinus rhythm in patients 1 and 3, and activation was seen to propagate from the endocardium to the epicardium. Electrocardiographic imaging was as effective at outlining the direction of activation (clockwise vs anticlockwise) as reconstructed endocardial data. However, the spread of activation differed at both levels when comparing isopotential and isochronal maps. Qualitatively, correlation was observed between regions of crowded isochrones at endocardial and epicardial level with 2 exceptions: when the pacing stimulus was at the septum and when regions of slow conduction were seen endocardially in the region of the septum. Reconstructed epicardial data were poor at determining activity in this region. Further quantitative analysis with integration of endocardial into reconstructed epicardial geometry was then performed. Table 1 shows the calculated distances between the first electrode to be activated using both systems. Comparative data for isochronal and isopotential maps are shown for all pacing sites. Epicardial isochronal maps were able to locate stimulation sites to a mean of 13.8 mm (SD, 4.7)
from corresponding endocardial data. Isopotential maps showed similar accuracy (mean, 14.0 mm; SD, 5.3). Despite the poor qualitative correlation during septal pacing, the interelectrode distances in this region were not significantly greater than during pacing of other regions (mean, 16.6 mm by isochrones and 13.8 mm by isopotentials). Epicardial isochronal maps were also able to locate regions of latest activation to a mean of 12.5 mm (SD, 3.7) from endocardial data (Table 1). Discussion This is the first clinical pacing study to compare epicardial data acquired noninvasively using ECGI with endocardial data acquired invasively using a noncontact mapping system. The data suggest that ECGI is as effective at locating LV endocardial pacing as invasive data except when the stimulus is located at a septal origin. Oster et al4 have shown that, by epicardial pacing of a dog heart in a human torso–shaped tank, single pacing sites over the epicardial surface could be reconstructed to within 10 mm or less of their measured
S116
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
positions. They also showed that regions of sparse and crowded isochrones could be accurately reconstructed using ECGI. Although problems with septal pacing have not previously been documented, there have been few data recorded in the clinical area. In addition, most studies have used epicardial pacing, making definitively septal stimulation difficult. Oster et al4 looked at anterior and posterolateral sites predominantly, whereas in a more recent study by McClelland et al,5 anterolateral, posterolateral, and posterior pacing sites were examined. Further work with pacing in all 5 segments documented in this study, particularly the septum, is undoubtedly required to substantiate these preliminary results. Further data are required to enable explanation of the discrepancy at the septum. One possibility relates to the structure and orientation of endocardium within the epicardial model. During pacing of the lateral wall, for example, the activation wavefront has only a short depth of myocardium to traverse before becoming apparent at epicardial level. Conversely, a pacing stimulus applied at the endocardial septum is further from all epicardial surfaces, let alone a well-defined septal wall on the 3-D geometry. It seems the activation wavefront could pass in 1 of 3 directions: (1) posteriorly to become apparent at the posterior epicardial wall, (2) towards the anterior wall, or (3) towards the right ventricle. In all 3 of our cases, the site of earliest activation was seen at the posterior epicardial wall. Interestingly, the endocardial wavefront in each case reached the posterior wall after leaving the septum, and there seemed to be a greater delay in initial epicardial activation (mean, 32.0 milliseconds). This compares to a mean delay of 8.5 milliseconds during pacing at alternative sites. Therefore, it might be the case that ECGI is unable to identify any activity at the septum, or alternatively, that septal activity is reconstructed with some delay at the nearest epicardial surface of the 3-D model. The attempt to quantitatively assess pacing site correlation has clear limitations. Anatomical segmentation into 5 regions was performed, not by surface area but according to regions defined by echocardiography. Therefore, the septum and apex, for example, represent a much smaller area than the posterior segment. This may explain why qualitatively, during pacing of the septum, the correlation seems poor, yet absolute interelectrode distances are not significantly greater than in other regions. By analyzing maps recorded in the absence of pacing, we were able to support the well-documented phenomenon that, in sinus rhythm, propagation of activation is mostly transmural, from endocardium to epicardium.3 Oster et al3 have previously shown that during pacing, the direction of the propagated activation wavefront varies according to stimulation depth and is dependent on myocardial fiber orientation. Again using an isolated dog heart suspended in a human torso tank, they demonstrated a clockwise rotation of the wavefront with subendocardial pacing and counterclockwise rotation with epicardial stimulation. From our study, performed in real time with human subjects and using endocardial/subendocardial pacing alone, we were unable to show clockwise rotation
exclusively for the activation wavefront. Only 6 of the 16 recordings supported these findings, with 2 cases showing predominantly counterclockwise rotation and the remaining 8 showing multidirectional propagation. We propose an important potential clinical application of ECGI is the anatomical localization of VT to aid ablation. Current long-term outcomes for VT ablation remain suboptimal (partly due to localization, but other variables including multiple sites and incomplete lesions transmurally) despite the advent of noncontact endocardial mapping, which has enabled even poorly tolerated arrhythmias to be treated.2 This is primarily due to the complexity of the VT circuit, a high proportion of which involves 3-D reentry18-20 or focal epicardial origin. Although the exit site has traditionally been targeted for ablation, recent work has shown the importance of locating the diastolic component of the circuit. Using the noncontact MEA in a clinical setting, Schilling et al7 were able to identify a complete tachycardia circuit within the diastolic interval in only 21%. In another study, this was identified in only 1 VT.2 One explanation is that at least part of the diastolic component is intramural or epicardial and therefore cannot be identified by the system.7 Burnes et al13 were the first to demonstrate the noninvasive reconstruction of reentrant activation at epicardial level during VT in a canine model. Combined acquisition of invasive endocardial data and reconstructed epicardial data would provide more comprehensive assessment of the VT diastolic circuit and exit site before radiofrequency ablation. Study limitations The main limitation of this study is the small numbers involved, with 16 recordings from only 3 patients. It is therefore with caution that we propose any definitive conclusion, rather emphasizing the need for further work. The limitations of ECGI are well documented, with torso inhomogeneities a major source of error.3 In addition, the inverse solution applied to BSPs used a general torso model rather than one tailored to individual body habitus. The ability to import patient-specific LV geometries by MRI/CT into the Prime reconstruct would be the obvious next step. Despite geometrical limitations, some studies have indicated that torso inhomogeneities affect only epicardial potential magnitudes and not isochronal patterns of activation.3,12 Finally, the use of reconstructed rather than directly measured data at epicardial and endocardial levels creates a further source of error. Conclusions This is the first study to support the use of ECGI in determining the site of endocardial pacing when compared directly with noncontact endocardial data. The small numbers employed in this study predict ECGI to be comparable to invasive endocardial data in all regions except during direct stimulation of the septum. Further data collection is essential before forming any definitive conclusion. We propose the combined use of ECGI and noncontact endocardial mapping to detail the complete diastolic circuit of VT, with the potential for guiding radiofrequency ablation of this difficult tachyarrhythmia.
H.J. Shannon et al. / Journal of Electrocardiology 40 (2007) S111–S117
References 1. Taccardi B, Arisi G, Macchi E, Baruffi S, Spaggiari S. A new intracavitary probe for detecting the site of origin of ectopic ventricular beats during one cardiac cycle. Circulation 1987;75:272. 2. Strickberger SA, Knight BP, Michaud GF, Pelosi F, Morady F. Mapping and ablation of ventricular tachycardia guided by virtual electrograms using a noncontact, computerized mapping system. J Am Coll Cardiol 2000;35:414. 3. Oster HS, Taccardi B, Lux RL, Ershler PR, Rudy Y. Electrocardiographic imaging: noninvasive characterization of intramural myocardial activation from inverse-reconstructed epicardial potentials and electrograms. Circulation 1998;97:1496. 4. Oster HS, Taccardi B, Lux RL, Ershler PR, Rudy Y. Non invasive electrocardiographic imaging: reconstruction of epicardial potentials, electrograms and isochrones and localization of single and multiple electrocardiac events. Circulation 1997;96:1012. 5. McClelland AJJ, Owens CG, Navarro C, et al. Usefulness of body surface maps to demonstrate ventricular activation patterns during left ventricular pacing and reentrant activation during ventricular tachycardia in men with coronary heart disease and left ventricular dysfunction. Am J Cardiol 2006;98:591. 6. Schilling RJ, Peters NS, Davies DW. Simultaneous endocardial mapping in the human left ventricle using a noncontact catheter. Circulation 1998;98:887. 7. Schilling RJ, Peters NS, Davies DW. Mapping and ablation of ventricular tachycardia with the aid of a noncontact mapping system. Heart 1999;81:570. 8. Gornick CC, Alder SW, Pederson B, Hauck J, Budd J, Schweitzer J. Validation of a new noncontact catheter system for electroanatomic mapping of left ventricular endocardium. Circulation 1999;99:829. 9. Khoury DS, Taccardi B, Lux RL, Ershler PR, Rudy Y. Reconstruction of endocardial potentials and activation sequences from intracavitary
10.
11.
12.
13. 14. 15.
16.
17.
18.
19.
20.
S117
probe measurements: localization of pacing sites and effects of myocardial structure. Circulation 1995;91:845. Khoury DS, Berrier KL, Badruddin SM, Zoghbi WA. Three-dimensional electrophysiological imaging of the intact canine left ventricle using a noncontact multielectrode cavitary probe: study of sinus, paced, and spontaneous premature beats. Circulation 1998;97:399. Green LS, Taccardi B, Ershler PR, Lux RL. Epicardial potential mapping: effects of conducting media on isopotential and isochrone distributions. Circulation 1991;84:2513. Rudy Y, Plonsey R. A comparison of volume conductor and source geometry effects on body surface and epicardial potentials. Circ Res 1980;46:283. Burnes JE, Taccardi B, Rudy Y. A noninvasive imaging modality for cardiac arrhythmias. Circulation 2000;102:2152. Ackerman MJ. The visible human project. Proc IEEE 1998;86:504. Navarro C, Anderson J, Allen J, Toro J, Riddell J, Adgey AAJ. Construction of a model of the torso for ECG problems based on CAD tools and images of the human body. Technol Health Care 2001;9:176. Yamashita Y, Takahashi T. Use of the finite element method to determine epicardial from body surface potentials under a realistic torso model. IEEE Trans Biomed Eng 1984;BME 31:611. Owens C, Navarro C, McClelland A, et al. Improved detection of acute myocardial infarction using a diagnostic algorithm based on calculated epicardial potentials. Int J Cardiol 2006;111:292. Pogwizd SM, Hoyt RH, Saffitz JE, Corr PB, Cox JL, Cain ME. Reentrant and focal mechanisms underlying ventricular tachycardia in the human heart. Circulation 1992;86:1872. Stevenson WG, Khan H, Sager P, et al. Identification of reentry circuit sites during catheter mapping and radiofrequency ablation of ventricular tachycardia late after myocardial infarction. Circulation 1993;88:1647. Kaltenbrunner W, Cardinal R, Dubuc M, et al. Epicardial and endocardial mapping of ventricular tachycardia in patients with myocardial infarction: is the origin of the tachycardia always subendocardially localized? Circulation 1991;84:1058.