Published online before print November 1, 2004, doi: 10.1161/01.CIR.0000147234.82755.90
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Circulation. 2004;110:3175-3180.)
© 2004 American Heart Association, Inc.
Arrhythmia/Electrophysiology |
Value of Noncontact Mapping for Identifying Left Ventricular Scar in an Ovine Model
From the Cardiology Department, Westmead Hospital, Westmead, Sydney, Australia.
Correspondence to Dr Pramesh Kovoor, Cardiology Department, Westmead Hospital, Corner Darcy and Hawkesbury Roads, Westmead, NSW 2145, Australia. E-mail kovoor{at}westgate.wh.usyd.edu.au
Received February 6, 2004; de novo received April 6, 2004; revision received June 8, 2004; accepted June 10, 2004.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
Background— We assessed the hypothesis that "virtual electrograms" from a noncontact mapping system (EnSite 3000) could be used to localize myocardial scar.
Methods and Results— Myocardial infarctions were induced in sheep by inflating an angioplasty balloon in the left anterior descending coronary artery for 3 hours. Scar mapping was performed on 8 sheep without inducible ventricular tachycardia by use of the noncontact mapping system and a 256-channel contact mapping system. Transmural mapping needles were inserted into myocardial regions that were (1) scarred, (2) peripheral to the scar, and (3) distant from the scar. Unipolar electrograms were exported from both systems and analyzed on a personal computer workstation. The percentage of myocardial scarring at each needle site was assessed histologically. Pearson’s correlation was used to assess the degree of association between various electrogram characteristics and the presence of myocardial scarring. The only noncontact electrogram characteristic that showed any association with the presence of myocardial scarring was the negative slope duration (contact, r=0.62, P<0.001; noncontact, r=0.23, P=0.004). The other electrogram characteristics studied were electrogram maximal deflection (contact, r=0.38, P<0.001; noncontact, r=0.03, P=0.75) and minimal slope (contact, r=0.42, P<0.001; noncontact, r=0.05, P=0.54).
Conclusions— Noncontact electrograms do not reliably identify ventricular scar. Alternative strategies such as importing computed tomography images into the geometry should be used when scar localization is important.
Key Words: electrophysiology • mapping • ablation • myocardial infarction • arrhythmia
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
A noncontact mapping system (EnSite 3000) is currently in use1–5 that calculates virtual electrograms simultaneously at 3360 locations within the chamber of interest, thus enabling arrhythmias to be mapped rapidly. The virtual electrograms computed by this noncontact mapping system have been shown to have a good morphological correlation with contact electrograms from the same location.1,2
Identifying ventricular scar is important because ventricular tachycardia (VT) reentrant circuits are often located in the border zone surrounding scarred myocardium.6–8 Accurate scar localization would also allow lines of block to be made between the scar and anatomic limits such as valve orifices.9 It is not clear whether noncontact virtual electrograms can be used to identify myocardial scarring. Pearson’s correlation, which is not sensitive to changes in signal amplitude, was used in previous studies to compare contact and noncontact electrograms. We therefore tested the hypothesis that the noncontact mapping system could be used to identify ventricular scar in an ovine model.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
Equipment
The EnSite 3000 system incorporates a multielectrode array (MEA) catheter, amplifier, and workstation that have previously been described in detail.2,4 In brief, the MEA consists of 64 wires, each with a laser-etched electrode, mounted on a 7.5-mL balloon. The electrical signals from the MEA are recorded by an amplifier with a sampling rate of 1.2 kHz and a filtering bandwidth of 0.1 to 300 Hz. Electrograms recorded from the MEA are analyzed by a boundary-element method to obtain the inverse solution to Laplace’s equation and to derive 3360 virtual endocardial electrograms. A mathematical constraint (spatial regularization) is placed on the final solution to reduce electrogram noise. These virtual electrograms are displayed on a workstation as color isopotential maps (Clarity Software, Endocardial Solutions).
The transmural mapping needles (manufactured in-house) had an external diameter of 1.1 mm and were 15 mm long. Each needle had 3 stainless steel ring electrodes (1.5 mm long) and a beveled stainless steel distal tip electrode separated by 3 Teflon spacers (1.5 mm long; see Figure 1). When inserted into the ventricle perpendicular to the epicardium, the most proximal ring electrode recorded subepicardial electrograms, the middle 2 electrodes recorded intramural electrograms, and the most distal electrode recorded endocardial electrograms. The contact electrograms from the transmural mapping needles were recorded on a 256-channel electrophysiologic mapping system (CardioMapp256, GE Medical Systems) as a unipolar signal with a reference electrode attached to the metal rib retractors. The CardioMapp256 amplifier had an input dynamic range of ±40 mV peak to peak, a sampling rate of 1 kHz, and a filtering bandwidth of 0.2 to 300 Hz.
|
Procedural Details
All procedures were performed in the Westmead Hospital vivarium with the approval of the hospital animal ethics committee. Myocardial infarctions (MIs) were induced by inflating a 3.0-mm angioplasty balloon in the middle of the ovine left anterior descending coronary artery equivalent for 3 hours as described by Reek et al.10 An electrophysiologic study was performed a median of 25 days later on 23 MI survivors to assess inducibility of VT. Sheep with inducible VTs were mapped in a different set of experiments.
Scar mapping was performed on 8 male sheep (weight range, 42 to 61 kg) with ventricular scarring but no inducible VT after MI. The mapping procedure was performed a median of 118 days (range, 27 to 189 days) after the MI. The sheep were sedated with 5% isoflurane via face mask and then anesthetized with an intravenous bolus of propofol (2 mg/kg) before intubation and ventilation. General anesthesia was maintained with 1% to 4% isoflurane in 100% oxygen, and saline (0.9% NaCl, 100 mL/h IV) was administered throughout the procedure. Angiographic sheaths were placed in the femoral arteries and veins. A left thoracotomy was performed with an incision through the fourth intercostal space. The ventricular scar was identified by inspection and palpation. Multielectrode mapping needles were inserted into the scar and scar periphery (within 2 cm of visible epicardial scar) with 1-cm spacing between needles. A range of 4 to 9 needles per sheep were inserted into visibly scarred regions, and 5 to 10 needles per sheep were inserted into the scar periphery. The remainder of the 20 needles used per case were inserted into normal myocardium at least 3 cm away from the scar (range, 5 to 9; see Figure 2).
|
An EnSite MEA was introduced via a retrograde aortic approach into the left ventricular apex. A roving electrophysiology catheter was used to acquire the geometry of the left ventricle. The geometry was then expanded to include the locations of all distal (endocardial) needle electrodes. A map marker identifying the location of each needle was also added to the geometry.
The MEA balloon was checked before and after the pacing studies to ensure that it had not been punctured by the transmural mapping needles. Bipolar pacing was performed from the right ventricular apex and all of the multielectrode needles via both the endocardial (e1+e2) and epicardial pair (e3+e4) of electrodes. Pacing was performed at twice the diastolic threshold at a cycle length of 400 ms. Electrograms were recorded simultaneously during all of the pacing studies.
The sheep was euthanatized after completion of the pacing studies, and a marker was sutured at the site of each mapping needle. The heart was then fixed in 10% formalin for 2 weeks. The block of myocardium surrounding each needle was excised, dehydrated with 100% ethanol, and embedded in paraffin wax. A 4-µm-thick section was cut from each block and stained with Gomori trichrome. The percentage of scar in each section was calculated by manually tracing the scar border on digital images with Scion Image software (Scion Image 1.63, Scion). In areas of myocardial thinning, the most distal needle electrode may have been located in the ventricular chamber and hence, measuring a cavity electrogram. We calculated the electrode (e1–e4) that would have been in contact with the endocardium by measuring the thickness of the myocardium surrounding each needle in the histologic sections and allowing for 10% shrinkage during tissue fixation. The electrode that was calculated to be subendocardial was used for all comparisons between the contact and noncontact electrograms.
From each pacing study, a grid of 256 virtual electrograms was exported from the EnSite system as an ASCII data file on a 250-MB Zip disk. The contact electrograms were exported from the CardioMapp256 system as binary files on a 100-MB Zip disk. A user-defined MatLab software program (version 5.3, The Mathworks) displayed and analyzed contact and noncontact electrograms simultaneously. Analysis was performed on the third paced beat. Both electrograms were adjusted to the same time point by alignment on the pacing spike. Because the noncontact mapping system sampled at a higher frequency (1.2 kHz) than the contact mapping system (1 kHz), every sixth value was discarded to display both electrograms to the same time scale. Analysis was performed on the 100-ms electrogram segment after the pacing spike.
The Matlab program was used to measure the voltage deflection by subtracting the lowest voltage value from the highest. The electrogram slope (first differential of the electrogram) was calculated for each value by subtracting the data point 2 ms before it from the data point 2 ms after it and dividing the result by 4. Hence, the electrogram slope that we obtained was the average slope over a 4-ms (5 data points) segment. The value and location of the most negative slope were calculated (minimum dV/dt). The duration of negative unipolar slope was calculated by measuring the time when the dV/dt was <20% of the most negative dV/dt (see Figure 3). Both sets of electrograms were then filtered with a 2-pole, Butterworth 4-Hz, high-pass filter. The filtered electrograms were then reanalyzed to calculate electrogram maximal deflection, minimum dV/dt, and negative slope duration as previously described.
|
We attempted to automate the "dynamic scar mapping" strategies that have been advocated as a possible method of detecting myocardial scar on virtual electrograms. These strategies are based on determining the pattern of activation during pacing from various sites near but outside of suspected scar. Scar is thought to be present in the areas bypassed by the activation wavefront (see Figure 1 in Reek et al11 for a graphic representation of this technique). Therefore, areas of scar would never have the lowest voltage compared with all of the electrograms at that time point. Because the scars were always in a similar location, human interpretation of location of scar from interactive analysis of wavefronts would have been biased. We therefore used an automated script to analyze the electrograms in the following manner. The initial 45 ms of ventricular activation was analyzed by a Matlab program. At each time point, each of the 256 virtual electrograms was ranked in comparison with the other 255, with a rank of 1 signifying the electrogram with the lowest voltage. The duration when the electrogram rank was in the lowest fifth, 15th, and 25th percentiles of ranks was then calculated. These values were calculated for unfiltered and for 4-, 8-, and 12-Hz high pass-filtered electrograms.
Statistical analysis was performed with SPSS for Windows (release 10.07, SPSS). Mean values for electrogram characteristics were obtained for each sheep to reduce the effect of repeated measurements in each sheep. Pearson’s correlation was used to calculate the correlation between the electrogram characteristics and the extent of histopathologically defined scarring at each site. The ability of contact and noncontact electrograms to identify areas of myocardial scarring (defined as >50% histopathologic scar) was assessed with receiver operator characteristic curves.
General linear models were fitted to the histopathologic percentage of scar. Two models were developed, one with the noncontact electrogram characteristics as covariates and the other with the contact electrogram characteristics. In each model, sheep identifier was included as a random factor. The proportion of total variability in percentage scar attributable to a particular variable was encapsulated in the eta2 term.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
Intraoperative scar identification was correlated well with subsequent histology. Areas identified as normal, peripheral to scar, and heavily scarred at surgery had 0.5±0.8%, 8.0±14.1%, and 54.6±27.3% mean percentage scar, respectively (r=0.77, P<0.001). Contact and noncontact electrograms were recorded a median of 36 mm from the MEA. The needle electrode that was calculated to be in contact with the endocardium was predominantly e1 (60%), followed by e4 (24%), then e2 (9%), and e3 (7%). In the scarred myocardium, e4 was calculated to be in contact with the endocardium in 57% of cases.
Contact endocardial electrogram characteristics were significantly correlated with the extent of scarring at that site. The best contact electrogram marker was the unfiltered negative slope duration (r=0.62, P<0.001), which was a mean of 12.5±4.3 ms in normal myocardium, 16.2±5.8 ms in the scar periphery, and 22.2±5.1 ms in scarred myocardium. The only noncontact electrogram characteristic that was significantly associated with the degree of scarring was the negative slope duration (r=0.23, P=0.004), which was 16.8±2.9 ms within normal myocardium, 17.6±3.3 ms within the scar periphery, and 18.8±3.5 ms within scarred myocardium. See Tables 1 and 2 
for the detailed results from unfiltered and 4-Hz high pass-filtered electrograms, respectively, and Figure 4. None of the dynamic analysis type of noncontact criteria were significantly correlated with ventricular scar. The highest correlation was for the duration when the ranked 12-Hz filtered electrogram was below the 25th percentile (r=0.155, P=0.051), but this was not significant.
|
|
|
Each of the contact electrogram parameters was independently significantly associated with scar in the fitted general linear model. The strongest association for contact electrograms was with the duration of negative slope (P<0.001, eta2=0.25), followed by bipolar electrogram size (P=0.004, eta2=0.06). The only significant independent predictors among the noncontact characteristics were minimum –dV/dt (P=0.02, eta2=0.04) and negative slope duration (P=0.04, eta2=0.03).
The clinical utility of contact and noncontact negative slope duration measurements for scar identification was analyzed by receiver operator characteristic curves. See Figure 5. For unfiltered contact electrograms, a cutoff value of 
16 ms gave a sensitivity of 94% and a specificity of 66%. The optimal cutoff value for unfiltered noncontact electrograms was 17 ms, which gave a sensitivity of 74% and a specificity of 47%.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
Previous studies investigating the accuracy of noncontact mapping in animal4 and clinical3 cases compared noncontact and contact electrograms with Pearson’s correlation. Because Pearson’s correlation is not affected by a change in scale (ie, amplitude) of the 2 data sets being measured,12 it is not clear whether noncontact electrograms can be used to identify myocardial scarring from electrogram voltage criteria. The use of data regularization strategies to reduce electrogram noise results in smoothing of the voltage differences between adjacent noncontact electrograms.13 This is the first study to determine whether noncontact electrograms can be used to identify histopathologically confirmed ventricular scarring. As expected, we found that contact electrogram voltage (unipolar and bipolar), minimal slope (dV/dt), and negative slope duration were correlated with the presence of histopathologically confirmed ventricular scarring. The contact unipolar electrogram voltage was reduced to 60% within scarred myocardium, which is similar to the levels reported in previous studies.14–16 There appeared to be no relation between the noncontact electrogram voltage (unipolar and bipolar) or minimal slope that could be used to predict the presence of ventricular scarring. Noncontact electrogram negative slope duration was significantly correlated with the extent of myocardial scarring, but this relation was much weaker than that for contact electrogram negative slope duration. The correlation was not improved when the analysis was repeated for only those noncontact electrograms recorded from within 40 mm of the MEA.
Reek and coworkers11 investigated the ability of noncontact mapping to identify left ventricular scar in 12 sheep after MIs were created in the left ventricular apex and septum. They found that the surface area of scar identified by noncontact unipolar voltage was correlated (r2=0.57) with the surface area of scar identified by magnetic resonance imaging. However, the slope of the correlation line was 3, indicating that there was poor agreement between the 2 data sets, despite the moderate correlation. Unfortunately, those authors were unable to coregister the positions of the electrograms with the location of ventricular scar, and therefore, they could not analyze whether the scar identified by the 2 systems was in the same location. The transmural mapping needles used in our study did not move once they had been pushed into the myocardium, allowing us to identify the specific location from which electrograms were recorded and therefore, correlate electrogram data with histopathologic findings. Callans et al16 performed a similar study with contact electrograms, measured on an electroanatomic mapping system, to define the size of left ventricular infarctions in 7 pigs with left anterior descending coronary artery territory infarctions. They found that the scar area defined by bipolar contact electrograms had a very high correlation (r2=0.97) and agreement (slope=1.03) with the scar area defined pathologically. Wrobleski et al17 provided further confirmation of the accuracy of contact mapping, finding a similar correlation (r2=0.94) between bipolar voltage mapping and pathologic scar. These latter investigators were able to determine that the location of the scar identified by the electroanatomic system was accurate by creating ablation lesions along its circumference.
We also attempted to determine whether dynamic scar mapping techniques were capable of detecting ventricular scar. We used an automated technique that prevented bias of interpretation. Our analysis showed no correlation between the dynamic mapping techniques and the presence of scar. This automated method of analysis is clearly less sophisticated than the interactive interpretation used by an expert operator using a digital workstation, but it should still be positive if the underlying theory is correct.
These findings have important clinical implications. Electrophysiologists should be aware that noncontact electrogram voltage cannot be used to map ventricular scarring. Alternative strategies such as using imported computed tomography cardiac or magnetic resonance images into the noncontact geometry are likely to provide this information in conjunction with activation maps.
In this experiment, we wished to identify criteria that could be used for automated scar identification. Such criteria need to be clearly defined numerically and independently of operator input. The negative slope duration is a simple numerical quantity that would be increased in the presence of fractionated electrical activity. The finding that the noncontact electrogram negative slope duration was significantly correlated with ventricular scarring is an important finding that may be useful for further research in noncontact scar mapping. The negative slope duration may be better at identifying scar because it is a temporal measurement rather than a voltage measurement and therefore, is less likely to be affected by spatial regularization. Previous studies have confirmed that electrogram duration criteria appear to offer some significant advantages over assessment of bipolar voltages alone for scar identification. Wrobleski et al17 found that 94% of sites that were incorrectly identified as scarred on voltage criteria alone (presumably owing to poor tissue contact at the recording site) could be correctly identified as normal by electrogram duration criteria. However, negative slope duration measured on standard noncontact electrograms does not have sufficient sensitivity and specificity to create accurate maps of ventricular scarring. Our data suggest that if noncontact electrograms are used to map the location of scar, they will have to be reconstructed with a different methodology. Alternative regularization strategies are available, and strategies incorporating temporal rather than spatial constraints may preserve the spatial electrogram changes associated with scarring.18 Additional processing beyond the standard boundary-element method may also improve the ability of noncontact mapping to identify scar.
|
Conclusions |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
Noncontact electrogram characteristics did not reliably detect ventricular scarring. In cases where scar localization is important, alternative strategies such as importing computed tomography images into the noncontact geometry should be used.
|
Acknowledgments |
|---|
Dr Thiagalingam was supported by a research scholarship from the Cardiac Society of Australia and New Zealand. This study was supported by grants from the Westmead Hospital Charitable Trust and the National Health and Medical Research Council of Australia awarded to Dr Kovoor. The authors would like to thank Dr Ross Mathews and the staff of the Westmead Hospital Animal Research Facility for their assistance.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
1. Strickberger SA, Knight BP, Michaud GF, et al. Mapping and ablation of ventricular tachycardia guided by virtual electrograms using a noncontact, computerized mapping system. J Am Coll Cardiol. 2000; 35: 414–421.
2. Schilling RJ, Davies DW, Peters NS. Characteristics of sinus rhythm electrograms at sites of ablation of ventricular tachycardia relative to all other sites: a noncontact mapping study of the entire left ventricle [see comments]. J Cardiovasc Electrophysiol. 1998; 9: 921–933.[Medline] [Order article via Infotrieve]
3. Schilling RJ, Peters NS, Davies DW. Simultaneous endocardial mapping in the human left ventricle using a noncontact catheter: comparison of contact and reconstructed electrograms during sinus rhythm. Circulation. 1998; 98: 887–898.
4. Gornick CC, Adler SW, Pederson B, et al. Validation of a new noncontact catheter system for electroanatomic mapping of left ventricular endocardium. Circulation. 1999; 99: 829–835.
5. Friedman PA, Asirvatham SJ, Grice S, et al. Noncontact mapping to guide ablation of right ventricular outflow tract tachycardia. J Am Coll Cardiol. 2002; 39: 1808–1812.
6. El-Sherif N, Hope RR, Scherlag BJ, et al. Re-entrant ventricular arrhythmias in the late myocardial infarction period, 2: patterns of initiation and termination of re-entry. Circulation. 1977; 55: 702–719.
7. Mehra R, Zeiler RH, Gough WB, et al. Reentrant ventricular arrhythmias in the late myocardial infarction period, 9: electrophysiologic-anatomic correlation of reentrant circuits. Circulation. 1983; 67: 11–24.
8. Wit AL, Allessie MA, Bonke FI, et al. Electrophysiologic mapping to determine the mechanism of experimental ventricular tachycardia initiated by premature impulses: experimental approach and initial results demonstrating reentrant excitation. Am J Cardiol. 1982; 49: 166–185.[CrossRef][Medline] [Order article via Infotrieve]
9. Marchlinski FE, Callans DJ, Gottlieb CD, et al. Linear ablation lesions for control of unmappable ventricular tachycardia in patients with ischemic and nonischemic cardiomyopathy. Circulation. 2000; 101: 1288–1296.
10. Reek S, Bicknell JL, Walcott GP, et al. Inducibility of sustained ventricular tachycardia in a closed-chest ovine model of myocardial infarction. Pacing Clin Electrophysiol. 1999; 22: 605–614.[CrossRef][Medline] [Order article via Infotrieve]
11. Reek S, Geller JC, Mittag A, et al. Noncontact mapping of ventricular tachycardia in a closed-chest animal model of chronic myocardial infarction. Pacing Clin Electrophysiol. 2003; 26: 2253–2263.[CrossRef][Medline] [Order article via Infotrieve]
12. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986; 1: 307–310.[CrossRef][Medline] [Order article via Infotrieve]
13. Gulrajani RM. The forward and inverse problems of electrocardiography. IEEE Eng Med Biol Magn. 1998; 17: 84–101, 122.[CrossRef]
14. Keck A, Hertting K, Schwartz Y, et al. Electromechanical mapping for determination of myocardial contractility and viability: a comparison with echocardiography, myocardial single-photon emission computed tomography, and positron emission tomography. J Am Coll Cardiol. 2002; 40: 1067–1074;discussion 1075–1078.
15. Koch KC, vom Dahl J, Wenderdel M, et al. Myocardial viability assessment by endocardial electroanatomic mapping: comparison with metabolic imaging and functional recovery after coronary revascularization. J Am Coll Cardiol. 2001; 38: 91–98.
16. Callans DJ, Ren JF, Michele J, et al. Electroanatomic left ventricular mapping in the porcine model of healed anterior myocardial infarction: correlation with intracardiac echocardiography and pathological analysis. Circulation. 1999; 100: 1744–1750.
17. Wrobleski D, Houghtaling C, Josephson ME, et al. Use of electrogram characteristics during sinus rhythm to delineate the endocardial scar in a porcine model of healed myocardial infarction. J Cardiovasc Electrophysiol. 2003; 14: 524–529.[CrossRef][Medline] [Order article via Infotrieve]
18. Velipasaoglu EO, Sun H, Zhang F, et al. Spatial regularization of the electrocardiographic inverse problem and its application to endocardial mapping. IEEE Trans Biomed Eng. 2000; 47: 327–337.[CrossRef][Medline] [Order article via Infotrieve]
This article has been cited by other articles:
 |
|
 |
|
  G. Sivagangabalan, J. Pouliopoulos, K. Huang, J. Lu, M. A. Barry, A. Thiagalingam, D. L. Ross, S. P. Thomas, and P. Kovoor Comparison of Electroanatomic Contact and Noncontact Mapping of Ventricular Scar in a Postinfarct Ovine Model With Intramural Needle Electrode Recording and Histological Validation Circ Arrhythm Electrophysiol, December 1, 2008; 1(5): 363 - 369. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Rao, Y. Ling, R. He, A. L. Gilbert, N. G. Frangogiannis, J. Wang, S. F. Nagueh, and D. S. Khoury Integrated multimodal-catheter imaging unveils principal relationships among ventricular electrical activity, anatomy, and function Am J Physiol Heart Circ Physiol, February 1, 2008; 294(2): H1002 - H1009. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. J. Abrams, M. J. Earley, S. C. Sporton, P. M. Kistler, M. A. Gatzoulis, M. J. Mullen, J. A. Till, S. Cullen, F. Walker, M. D. Lowe, et al. Comparison of Noncontact and Electroanatomic Mapping to Identify Scar and Arrhythmia Late After the Fontan Procedure Circulation, April 3, 2007; 115(13): 1738 - 1746. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||