Simultaneous Endocardial Mapping in the Human Left Ventricle Using a Noncontact Catheter
Comparison of Contact and Reconstructed Electrograms During Sinus Rhythm
Background—Catheter ablation of ventricular tachycardia is limited in part by difficulty in identifying suitable sites for ablation. A noncontact multielectrode array (MEA) has been developed that allows reconstruction of 3360 electrograms, using inverse-solution mathematics, that are superimposed onto a computer-simulated model of the endocardium. This study assesses the accuracy of timing and morphology of reconstructed unipolar electrograms compared with contact unipolar electrograms from the same endocardial site.
Methods and Results—The MEA was deployed in the left ventricles of 13 patients (end-diastolic diameters, 61.7±8.4 mm [mean±SD]). We recorded contact electrograms at 76 points equatorial and 32 points nonequatorial to the MEA during sinus rhythm using a catheter-locator signal to record direction and distance from the MEA. Morphology (cross-correlation) and timing of maximum −dV/dt of contact and reconstructed electrograms were compared at different distances from the MEA center to endocardium (M-E) and from the MEA equatorial plane. For equatorial data, the M-E was 32.12±12.12 mm. The timing of reconstructed with respect to contact electrograms was −1.94±7.12 ms for M-E <34 mm and −14.16±19.29 ms at M-E >34 mm (P<0.001). Cross-correlation of electrograms was 0.87±0.12 (95% CI, 0.84 to 0.91) and 0.76±0.18 (95% CI, 0.69 to 0.83) for M-E <34 mm and >34 mm, respectively. Nonequatorial points were 32.33±10.81 mm (range, 16.9 to 55.6 mm) from the MEA equatorial plane; electrogram timing difference was −8.97±15.75 ms and was unrelated to this distance from the equator.
Conclusions—This noncontact mapping system accurately reconstructs endocardial unipolar electrograms from the human left ventricle. At M-E distances >34 mm, timing accuracy of reconstruction decreases.
Reentry is the mechanism responsible for the majority of cases of ventricular tachycardia (VT).1 2 3 4 5 Ablation of VT is dependent on locating the diastolic activity critical for maintenance of the reentrant circuit.5 6 Only 10% of patients with structural heart disease and reentrant VT are suitable for catheter ablation by use of conventional techniques7 ; the principal reason for this limitation is hemodynamic intolerance of the tachycardia for a sufficient time to allow conventional sequential endocardial mapping.
Simultaneous endocardial activation mapping with basket electrodes has been achieved with limited resolution.8 More detailed simultaneous activation mapping has been performed during surgery by applying arrays of multiple electrodes to epicardial or endocardial surfaces.9 10 11 This method of mapping requires a thoracotomy and general anesthesia, and many VTs may be nonsustained or noninducible under these conditions.12 There has therefore been no means of producing high-resolution activation maps of the entire intact left ventricle (LV) that would enable adequate mapping of the tachycardia circuit in just a few beats and thereby potentially guide ablative therapy for the large proportion of patients with poorly tolerated VT.
We report the first human studies using a noncontact mapping system to detect far-field endocardial potentials from within a cardiac chamber and, from these potentials, reconstruct unipolar endocardial electrograms at 3360 points, producing instantaneous endocardial isopotential maps on a computer-generated “virtual” endocardium. The aim of this study was to validate the system during sinus rhythm (SR) by comparing the timing and morphology of reconstructed unipolar electrograms with contact unipolar electrograms recorded from the same LV endocardial location.
Thirteen consecutive patients (mean age, 60 years; range, 32 to 75 years; 1 woman) were studied while undergoing endocardial LV mapping for catheter ablation of well-tolerated VT (Table 1⇓). All patients were being treated with amiodarone therapy, and other antiarrhythmic medications were continued for the study.
The study was approved by the local ethics committee, whose guidelines were followed. All patients were studied in the postabsorptive state and had given written informed consent. A standard quadripolar catheter was sited in the right ventricular apex, and 2 standard, deflectable, mapping/ablation, 4-mm-tip catheters were passed to the LV, 1 by a retrograde transaortic route and the other via a transeptal puncture. Pulmonary and systemic arterial blood pressures were monitored continuously.
Noncontact Mapping System
The noncontact mapping system (EnSite 3000; Endocardial Solutions) consists of a catheter-mounted multielectrode array (MEA), a custom-built amplifier system, and a Silicon Graphics workstation to run specially designed system software.
The MEA (a woven braid of 64 0.003-in-diameter wires) is mounted on a 7.6-mL balloon on a 9F catheter (Figure 1⇓). Each wire has a 0.025-in break in insulation, producing a noncontact unipolar electrode. The raw far-field electrographic data from the MEA are acquired and fed into a multichannel recorder and amplifier system, sampled at 1.2 kHz, and filtered with a bandwidth of 0.1 to 300 Hz. The amplifier also has 16 channels for contact catheters and 12 for the surface ECG. A ring electrode located on the proximal shaft of the MEA catheter in the descending aorta is used as a reference for both noncontact and contact unipolar electrogram recordings.
Before deployment of the MEA, patients were given 10 000 IU heparin with later boluses to maintain activated clotting time at 300 to 400 seconds. The MEA catheter was deployed via the retrograde transaortic route over a 0.032-in J-tipped guidewire advanced to the LV apex. With the pigtail of the MEA in the LV apex, the guidewire was withdrawn and the balloon inflated with a contrast-saline mixture (Figure 2⇓).
The system is able to locate any conventional catheter in space with respect to the MEA by passing a 5.68-kHz, low-current “locator” signal between the catheter being located and alternately between ring electrodes proximal and distal to the MEA on the noncontact catheter. The MEA detects and determines the locator signal angles and thus positions the source.
This locator signal serves 2 purposes. First, it is used to provide measured samples for the geometry matrix of the inverse solution by constructing a 3-dimensional computer model of the endocardium (virtual endocardium). This was achieved in the present study either by use of an elliptical model of the LV fitted to 6 endocardial points equatorial to the balloon or by dragging a mapping catheter around the LV chamber, building up a series of coordinates for the endocardium (contour geometry), generating an anatomically more accurate endocardial model. Geometric points were sampled at the beginning of the study during SR and were taken by gating 6 ms before the R wave of the ECG. The locator signal is also used to display and log the position of the mapping catheter on the virtual endocardium during a study (Figure 3⇓).
Reconstruction of Electrograms
The electrical activity detected by the MEA is generated primarily by the electrograms on the endocardial surface and is of lower amplitude and frequency than the source on the endocardium. A technique to enhance and resolve these far-field potentials has been devised based on an inverse solution to Laplace’s equation by use of a boundary element method (BEM). The inverse solution considers how a signal detected at a remote point will have appeared at the source, and the BEM is a method for applying the inverse solution to resolve a matrix of such signals from a source at a known boundary (eg, the blood-endocardial boundary).
The potential distribution on the MEA created by potentials at the blood-endocardial boundary is described by Laplace’s equation. The potential field at any 1 electrode is influenced to a degree by the potentials from the entire endocardium, the degree of influence being inversely proportional to the distance between the electrode and each endocardial point. The potential field created on the MEA surface is therefore related to the MEA–endocardial geometry matrix. When this is known, it is possible to compute endocardial electrograms from the MEA potentials by inverse solution of Laplace’s equation.
The inverse solution is inherently ill posed, meaning that noise from the MEA electrodes or inaccuracy in the MEA–endocardial geometry matrix results in large errors in reconstruction of electrograms. To minimize this, stability is provided by application of a mathematical constraint, with physiological basis, to the solution by use of a technique called regularization,13 14 for which a custom-designed algorithm based on methods described by Tikhonov and Arsenin14 is used. Accuracy of reconstruction of endocardial electrograms is therefore dependent on the solution to Laplace’s equa-tion, the regularization technique used, and the accuracy of the geometry matrix. Errors in geometry will still occur and may be related to the number of endocardial points sampled and the complexity of the geometry of the chamber. These errors will be reflected in the results shown later.
Several modifications have been applied to previously described techniques for inverse solution15 to improve the accuracy of the system. The inverse solution is based on Green’s second formula and is executed by use of higher-order algebraic expressions:
where D is a domain, ∂D is the boundary of D, ∂/∂ n represents the outward normal on D, ∇2 is the Laplacian, dA is the surface area differential, dD is the volume differential, w is the potential field created by a unit charge in free space, and v is a solution of the Laplace equation.
The geometry matrix defines the relationship between the location of the 64 electrodes on the MEA and 3360 points on the endocardium where the reconstruction is computed. Theoretically, the use of a model based on linear splines may cause significant errors in the geometry estimation because the sharp pyramidal points of the spline are a poor estimate of the curvature of the endocardial surface. Therefore, a model based on bicubic splines, which fits the sampled endocardial points with curves rather than lines, was used.
With these techniques, the system can, from the MEA potentials, reconstruct and interpolate >3300 electrograms simultaneously over the entire virtual LV endocardium (Figure 3⇑). The use of a BEM by the inverse solution means that the 3-dimensional myocardium is treated as a 2-dimensional endocardial surface as it is by contact endocardial catheters. Also, reconstructed electrograms are subject to the same electrical principles as contact electrograms, so that both contain contributions from surrounding endocardium as well as underlying myocardium. The contribution of each component to the electrogram is weighted by amplitude and distance from the site of measurement. In generating the geometry matrix, data from structures protruding into the chamber cavity are likely to be ignored in favor of data from more distant endocardial points on the same vector. Structures such as papillary muscles, therefore, are not likely to cause significant distortion of the virtual endocardium but, as with local contact electrograms, may make a small contribution to the far-field electrograms.
Contact catheter data and the surface ECG were recorded simultaneously on a conventional electrophysiology system (Bard Labsystem or Prucka Cardiolab) and the noncontact mapping system. Both the contact electrograms and the position of the contact catheter as determined by the locator signal were recorded simultaneously. Reconstructed electrograms could be selected later at any location on the virtual endocardium and displayed individually, allowing comparison of a contact electrogram and a reconstructed electrogram from the same endocardial location.
Validation of Electrogram Reconstruction
Contact unipolar recordings from a mapping catheter were made in all patients during SR from ≥6 separate endocardial locations equatorial to the balloon. The accuracy of reconstruction was compared with the distance from the MEA center as defined by the catheter-location system, which has been validated by in vitro and animal studies.16 The perpendicular equatorial distance of the contact catheter from the true equatorial plane of the balloon was calculated by the catheter-location system and displayed in real time. Equatorial positions were defined as endocardial sites within 15 mm perpendicular to the true equatorial plane of the MEA. Recordings were also made at points >15 mm from the MEA equator to assess the accuracy of reproduction toward the poles of the MEA. Contact catheter sites were taken from as varied an anatomic distribution as possible for each patient (Figure 4⇓).
The effects of catheter pressure on local electrogram timing and morphology have been demonstrated.17 Therefore, unipolar electrograms from the ring electrode, 2 mm from the mapping catheter tip, were also compared with the reconstructed electrogram to explore any differences resulting from pressure.
All locations were marked by use of the locator signal. Software filtering produced a bandwidth of 4 to 300 Hz for both contact and reconstructed electrograms.
Data were analyzed on the Silicon Graphics workstation as follows:
1. Electrogram timings were compared at the point of maximum −dV/dt in the reconstructed and contact electrograms.
2. Differences in electrogram morphology, with reference to polarity, relative amplitude, and frequency of electrogram components, were compared after the gains had been adjusted to produce an optimal visual match. Electrogram morphologies were examined by overlaying reconstructed and contact electrograms recorded from the same site, and a visual morphology score (1 to 5) was used wherein 5 indicates an exact match (no difference could be seen in overlying electrograms), 4 indicates features are exact with phase shift (no visible difference in electrogram morphology, with an overall shift in timing), 3 indicates many features match with or without phase shift (the overlying electrograms match closely, but differences in electrogram features can be seen), 2 indicates few features match with or without phase shift (similar features can be identified in the electrograms but with a poor overall match), and 1 indicates no match (<2 clearly identifiable features can be matched).
4. Off-line cross-correlation can be used to measure differences in timing between reconstructed and contact catheter electrograms. A computer shifts the reconstructed against the contact electrograms in time to produce the maximum correlation between the 2 sets of amplitude samples, and this timing shift is called correlation timing.
Continuous data, including differences in timing, correlation, and endocardial distance, are presented as means, SDs, and 95% CIs. Means of continuous data were compared with Student’s t test. Negative values for comparison of timing indicate that the reconstructed electrograms had earlier maximum −dV/dt than their contact counterparts. Results of the noncontinuous subjective morphology score were compared with endocardial distances by use of general linear model ANOVA and linear regression analysis.
Results are summarized in Table 2⇓.
Data were collected from 76 equatorial LV points (mean, 5.9 points/patient; range, 5 to 7).
Timing. Perfect matches in timing were obtained as far as 52 mm from the MEA center, but differences in timing of maximum −dV/dt for reconstructed electrograms with respect to contact counterparts increased gradually with distance and significantly at data points >34 mm from the MEA center (Figure 5A⇓). Thus, data were handled in 2 sets of points <34 mm and >34 mm from the center of the MEA. Computer timing analysis did not demonstrate this predominantly negative shift of reconstructed electrograms with respect to contact electrograms (Figure 5B⇓).
Morphology. The subjective morphology score (Figure 6A⇓) showed no apparent threshold distance beyond which the morphology of reconstructed electrograms worsened (Figure 7A⇓). Cross-correlation worsened at greater endocardial distances, but a threshold distance for this was not clear (Figure 7B⇓).
Thirty-two nonequatorial points were recorded (mean, 2.5 points/patient; range, 1 to 6). The mean perpendicular distance of points from the equatorial plane of the MEA for nonequatorial data was 32.33±10.81 mm (range, 16.9 to 55.6 mm).
Timing. Perfect electrogram reconstruction was demonstrated at distances of up to 44.1 mm perpendicular to the MEA equator. No relationship was demonstrated between perpendicular distance from the equator and timing differences (Figure 8A⇓).
Morphology. The mean morphology score of nonequatorial reconstructed electrograms (Figure 6B⇑) showed no statistically significant trend to alter with increasing perpendicular distance from the equator (Figure 8B⇑).
The commonest cause for VT in coronary heart disease is reentry.21 22 The ability to ablate such VT with catheters has been limited by the time taken to adequately map endocardial activation with conventional sequential techniques, restricting this therapy to a subset of patients with hemodynamically stable VT.
The efficacy of catheter ablation of VT has been disappointing. Although 71% to 90%5 7 23 24 25 26 of VTs are rendered noninducible, long-term recurrence rates are high. Attempts to improve these results have been directed both at increasing the size of ablation lesions and at improving mapping techniques to increase the precision of delivery of lesions to the complex substrate causing reentry in ischemic VT.
Detailed simultaneous endocardial and epicardial activation mapping of VT has been performed at surgery by application of arrays of multiple electrodes directly to the epicardium or endocardium.10 However, mapping under anesthetic with an open chest and possible ventriculotomy has an associated morbidity and mortality.27 28 Simultaneous mapping of multiple points can also be performed by use of endocardial basket arrays deployed percutaneously. Resolution remains limited to the proportion of electrodes in contact with the endocardium and by unequal deployment and spacing of the splines.8
Three-dimensional electroanatomic reconstruction of sequentially acquired contact catheter data has also recently been described and validated.20 Although this is a significant advance in mapping technology, like all sequential systems, resolution is limited by the time available to acquire separate data points, and its use remains restricted in nonsustained or hemodynamically unstable arrhythmias.
Noncontact mapping has several potential advantages over these techniques because high-resolution maps of the entire intact cardiac chamber are created simultaneously, producing complete maps of each beat of tachycardia. Noncontact endocardial mapping was first described by Taccardi et al29 when olive-shaped and cylindrical probes were used to recorded noncontact electrograms of ventricular-paced beats in dogs. Such recordings resulted in low-frequency, low-amplitude cavity potentials.30 To enhance this technique, principles that had previously been applied to reconstruction of epicardial maps from skin-surface electrograms31 were applied to reconstruct endocardial electrograms.15 32 33 Endocardial reconstruction has been applied by use of cylindrical probes in an open chest dog with epicardial echocardiography used to provide a geometry matrix, with good correlation between contact and reconstructed electrograms.15
The system reported herein combines inverse-solution mathematics with a catheter-location device and thus has potential advantages over previous noncontact systems. Cavity geometry is defined with limited need for an alternative imaging system, and the progress of a mapping catheter within the cavity can be monitored continuously and recorded, potentially reducing exposure to x-rays.
This system was first validated in vitro34 when accurate catheter location and good correlation were seen between reconstructed and contact potentials by use of a lower-order inverse solution than presented in this article. With the presently used inverse solution, reconstruction of electrograms and catheter-location accuracy decreased at distances >40 mm from the MEA center.16 In normal dogs, cross-correlation between reconstructed and contact unipolar electrograms was 0.98, and a mapping catheter was positioned to within 4.0±3.2 mm of a target pacing plunge electrode.16 The LVs of those dogs were smaller (3.98±0.22 cm LV end-diastolic dimension [mean±SD]) than the LVs of the patients in the present study (6.17±0.84 cm), which were typical of patients with ischemic heart disease and VT.
The results of the present study show that this system can accurately reconstruct electrograms at distances of at least 50 mm from the center of the MEA but that accuracy decreases with distance, significantly so >34 mm from the MEA center. There is no significant difference in timing of maximum −dV/dt between contact catheter and reconstructed electrograms <34 mm from the MEA center, but the timing of reconstructed electrogram maximum −dV/dt becomes earlier than that of contact electrograms >34 mm from the MEA center. The trend for the reconstructed electrograms to be displayed earlier than contact electrograms at distances >34 mm from the MEA center probably results from a combination of errors. These may be introduced by distance of the endocardium from the MEA, signal-to-noise ratio, complexity of anatomy, motion of different regions of the LV, and the presence of scar. These possible sources of error may be further compounded by regularization, in which subtle features of the endocardial potentials may be rejected because of rapidly changing geometry or electrogram amplitude. Thus, the complex relationship between reconstruction errors and MEA-endocardial distance can be easily demonstrated but not easily defined. If this complex, reproducible error in the reconstruction algorithm is defined, it may be possible to modify the mathematical solution to adjust for this.
The computerized correlation timing data also demonstrated that differences in electrogram timing increased with increasing endocardium-to-MEA distances, but there was not a consistent shift in timing of reconstructed electrograms to precede contact electrograms, and differences in timing were less than those measured by comparison of maximum −dV/dt. Computerized correlation timing provides a more global measurement of electrogram timing but may underestimate errors in activation timing of reconstructed electrograms and requires further validation.
Previous data have suggested that accuracy of electrogram reconstruction decreases at points toward the poles of the MEA.16 Our data show no such trend, and although the morphological accuracy of electrogram reconstruction decreased with increasing distance from the center of the MEA, the morphology did not alter significantly with points at increasing perpendicular distances from the MEA equator. This may be explained by the ellipsoid shape of the balloon, because although points at increasing angles from the MEA equator are farther from the MEA center, they may be closer to the polar surface of the array.
There was no difference between the results of tip- and ring-electrode electrogram comparisons, which suggests that the pressure effect was not significant in this study.
Limitations of Study
The design of this study in humans meant that the display of reconstructed electrograms for comparison with contact electrograms depended on both the reconstruction algorithm and the accuracy of the catheter-location system. This study is limited by its inability to determine whether errors seen were due to the reconstruction process or the location system. However, although previous in vitro data have supported the accuracy of the locator system and construction of the virtual endocardium,16 no animal study has yet compared the virtual endocardium with the true anatomy of the cardiac chamber. Because this study was performed on humans, it was not possible to precisely locate the catheter by other techniques and thus independently determine the accuracy of the locator system. This mapping system integrates the locator and electrogram reconstruction technologies, and this study addresses the performance of the system as a whole. Because the principal purpose of the mapping system described herein is to map human VT, further study is required to confirm that data obtained during VT, particularly small-amplitude diastolic potentials, can also be accurately reconstructed. A theoretical limitation is that geometry data are acquired during SR but applied to mapping of VT, when the LV dimensions may be different.
The noncontact mapping system described in this study can faithfully reconstruct endocardial electrograms in the human LV during SR. The accuracy of reconstruction depends on the distance between the endocardium and the center of the MEA. The consistent nature of the timing error suggests that the reconstruction of electrograms at MEA-endocardial distances >34 mm may be improved with refinements in the mathematical solution and hardware used. Further studies will determine whether this system can assist in the mapping and treatment of human arrhythmias.
Richard Schilling is a British Heart Foundation Junior Research Fellow. The authors would like to thank Dr Gerry C. Kaye of Castle Hill Hospital, Hull, UK, for his assistance with this study.
- Received December 2, 1997.
- Revision received April 1, 1998.
- Accepted April 20, 1998.
- Copyright © 1998 by American Heart Association
Josephson ME, Horowitz LN, Farshidi A, Kastor JA. Recurrent sustained ventricular tachycardia, I: mechanisms. Circulation. 1978;57:431–439.
Karagueuzian HS, Fenoglio JJ, Weiss MB, Wit AL. Protracted ventricular tachycardia induced by premature stimulation in the canine heart after coronary artery occlusion and reperfusion. Circ Res. 1979;44:833–846.
Wellens HJJ, Schuilenberg RM, Durrer D. Electrical stimulation of the heart in patients with ventricular tachycardia. Circulation. 1972;46:216–226.
Stevenson WG, Khan H, Sager P, Saxon LA, Middlekauff HR, Natterson PD, Wiener I. Identification of reentry circuit sites during catheter mapping and radiofrequency ablation of ventricular tachycardia late after myocardial infarction. Circulation. 1993;88:1647–1670.
El-Sherif N, Mehra R, Gough WB, Zeiler RH. Reentrant ventricular arrhythmias in the late myocardial infarction period: interruption of reentrant circuits by cryothermal techniques. Circulation. 1983;68:644–656.
Morady F, Harvey M, Kalbfleisch SJ, El-Atassi R, Calkins H, Lanberg JJ. Radiofrequency ablation of ventricular tachycardia in patients with coronary artery disease. Circulation. 1993;87:363–372.
Johnson SB, Packer DL. Intracardiac ultrasound guidance of multipolar atrial and ventricular mapping basket applications. J Am Coll Cardiol. 1997;29:202A. Abstract.
Downar E, Saito J, Doig JC, Chen TC, Sevaptsidis E, Masse S, Kimber S, Mickleborough L, Harris L. Endocardial mapping of ventricular tachycardia in the intact human ventricle, III: evidence of multiuse reentry with spontaneous and induced block in portions of reentrant path complex. J Am Coll Cardiol. 1995;25:1591–1600.
Tikhonov AN, Arsenin VY. Solutions of Ill-Posed Problems. Washington, DC: VH Winston & Sons; 1977:27–94.
Khoury DS, Taccardi B, Lux RL, Ershler PR, Rudy Y. Reconstruction of endocardial potentials and activation sequences from intracavity probe measurements: localization of pacing sites and effects of myocardial structure. Circulation. 1995;91:845–863.
Gornick CC, Adler SW, Pederson B, Hauck J, Budd J, Schweitzer J. Validation of a new noncontact catheter system for electroanatomic mapping of the left ventricular endocardium. Circulation. In press.
Gepstein L, Hayam G, Ben-Haim S. A novel method for nonfluoroscopic catheter-based electroanatomical mapping of the heart. Circulation. 1997;95:1611–1622.
Myerburg RJ, Kessler KM, Kimura S, Castellanos A. Sudden cardiac death: future approaches based on identification and control of transient risk factors. J Cardiovasc Electrophysiol. 1992;3:626–640.
Kadish A, Morady F. Mechanism-specific therapy of ventricular tachycardia. J Cardiovasc Electrophysiol. 1990;1:68–69.
Rothman SA, Hsia HH, Cossú SF, Chmielewski IL, Buxton AE, Miller JM. Radiofrequency catheter ablation of postinfarction ventricular tachycardia: long-term success and the significance of inducible nonclinical arrhythmias. Circulation. 1997;96:3499–3508.
Borgreffe M, Chen X, Hindricks G, Haverkamp W, Willems S, Martinez-Rubio A, Shenasa M, Rotman B, Breithardt G. Catheter ablation of ventricular tachycardia in patients with heart disease. In: Zipes DP, ed. Catheter Ablation of Arrhythmias. Armonk, NY: Futura Publishing Co Inc; 1994:277–308.
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;75:272–281.
Khoury DS, Rudy Y. A model study of volume conductor effects on endocardial and intracavitary potentials. Circ Res. 1992;71: 511–525.
Rudy Y, Messinger-Rapport BJ. The inverse problem in electrocardiography: solutions in terms of epicardial potentials. Crit Rev Biomed Eng. 1988;16:214–268.
Khoury DS, Rudy Y. Reconstruction of endocardial potentials from intracavitary probe potentials: a model study. Proceedings of Computers in Cardiology, Durham, NC, Oct 11–14, 1992. Piscataway, NJ: The Institute of Electrical and Electronics Engineers Inc Service Center; 1992.
Macchi E, Arisi G, Colli-Franzone P, Guerri L, Olivetti G, Taccardi B. Localization of ventricular ectopic beats from intracavitary potential distributions: an inverse model in terms of sources. Proceedings of the 11th Institute of Electrical and Electronics Engineers Engineering in Medicine and Biology Society, Piscataway, NJ: Institute of Electrical and Electronics Engineers Inc Service Center; 1989.