A Novel Method for Nonfluoroscopic Catheter-Based Electroanatomical Mapping of the Heart
In Vitro and In Vivo Accuracy Results
Background Cardiac mapping is essential for understanding the mechanisms of arrhythmias and for directing curative procedures. A major limitation of the current methods is the inability to accurately relate local electrograms to their spatial orientation. The objective of this study was to present and test the accuracy of a new method for nonfluoroscopic, catheter-based, endocardial mapping.
Methods and Results The method is based on using a new locatable catheter connected to an endocardial mapping and navigating system. The system uses magnetic technology to accurately determine the location and orientation of the catheter and simultaneously records the local electrogram from its tip. By sampling a plurality of endocardial sites, the system reconstructs the three-dimensional geometry of the chamber, with the electrophysiological information color-coded and superimposed on the anatomy. The accuracy of the system was tested in both in vitro and in vivo studies and was found to be highly reproducible (SD, 0.16±0.02 [mean±SEM] and 0.74±0.13 mm) and accurate (mean errors, 0.42±0.05 and 0.73±0.03 mm). In further studies, electroanatomical mapping of the cardiac chambers was performed in 34 pigs. Both the geometry and activation sequence were repeatable in all pigs.
Conclusions The new mapping method is highly accurate and reproducible. The ability to combine electrophysiological and spatial information provides a unique tool for both research and clinical electrophysiology. Consequently, the main shortcomings of conventional mapping—namely, prolonged x-ray exposure, low spatial resolution, and the inability to accurately navigate to a predefined site—can all be overcome with this new method.
Cardiac mapping was reported as early as 19151 and implies the registration of the electrical activation sequence by recording of extracellular electrograms. One objective of cardiac mapping is to analyze the activation wave fronts emerging from those structures that are involved in the genesis of arrhythmias.2 The exact localization of such structures is a prerequisite for understanding the pathophysiological mechanisms that underlie the arrhythmia, evaluating the effect of drugs,3 and directing surgical2 4 or catheter-ablation5 6 7 procedures.
Cardiac mapping is a broad term that covers several modes of mapping such as body-surface,8 endocardial,2 9 and epicardial10 mapping. Nevertheless, common to all cardiac mapping procedures is the registration of the electrical activation sequence by recording of extracellular electrograms. Electrogram acquisition can be simultaneous from many sites, usually during surgery, with the use of fixed-shape electrode arrays11 12 13 14 such as epicardial socks and endocardial balloons.9 15 More commonly, cardiac mapping is performed with catheters that are introduced percutaneously into the heart chambers and sequentially record the endocardial electrograms. These electrophysiological catheters are navigated and localized with the use of multiplanar fluoroscopy. A major pitfall in the currently used methods is the inability to accurately associate the intracardiac electrogram with a specific endocardial site. Thus, the localization of the recording sites with fluoroscopy is inaccurate, cumbersome, and associated with high x-ray exposure for both the patient and the physician.
We present here a new method for nonfluoroscopic catheter-based endocardial mapping that enables the generation of 3D electroanatomical maps of the heart chambers. This method is based on using a new locatable catheter connected to an endocardial mapping and navigation system. The system comprises a miniature passive magnetic field sensor located at the tip of the catheter, an external ultralow magnetic field emitter, and a processing unit. The system uses the magnetic technology to accurately determine the location and orientation of the catheter in six degrees of freedom (x, y, z, roll, pitch, and yaw) and simultaneously records the intracardiac local electrogram from its tip. The 3D geometry of the chamber is reconstructed in real time with the electrophysiological information, which is color-coded, superimposed on the electroanatomical map. The ability to localize the catheter tip in real time with reference to the electroanatomical map has special value for guiding catheter ablation procedures in which radiofrequency energy is delivered from the tip. For this application, it is essential that the navigation tools allow accurate and reproducible repositioning of the catheter tip on top of a previously defined target.
The present study describes the mapping technology and provides both in vitro and in vivo accuracy results of the navigation system. We also present typical electroanatomical maps of cardiac chambers of the swine during different rhythms. The possible contribution of this methodology to both clinical and research electrophysiology is discussed.
The mapping and navigation system comprises a miniature passive magnetic field sensor, an external ultralow magnetic field emitter (location pad), and a processing unit (CARTO, Biosense).
The locatable catheter is similar to a regular electrophysiological 8F deflectable-tip catheter (Cordis-Webster; Fig 1A⇓). The catheter tip is mounted on the distal end of the shaft and includes the tip electrode and several additional proximal electrodes that enable recording of unipolar or bipolar signals. Just proximal to the tip electrode lies the location sensor, totally embedded within the catheter. Signals received within the sensor are transmitted along the catheter shaft to the main processing unit. The catheter is equipped with radiofrequency delivery capabilities, including a 4-mm tip and a thermocouple, and can be used with an ordinary radiofrequency generator.
The locator pad is located beneath the operating table and generates ultralow magnetic fields (5×10−6 to 5×10−5 T) that code the mapping space around the patient’s chest with both temporal and spatial distinguishing characteristics. These fields contain the information necessary to resolve the location and orientation of the sensor in six degrees of freedom (x, y, z, roll, pitch, and yaw). The locator pad includes three coils (Fig 1B⇑). Each coil generates a magnetic field that decays as a function of the distance from that coil. The sensor measures the strength of the magnetic field, thus enabling determination of the distance from each of its sources. These distances determine the radii of theoretical spheres around each coil. The intersection of these three spheres determines the location of the sensor in space.
Two locatable 8F catheters were introduced, by use of fluoroscopic guidance, through the appropriate vascular access and positioned inside the heart chambers. The first catheter was introduced and placed in the CS or the RVA and served as a reference catheter. The mapping catheter was introduced and placed in the mapped chamber. The mapping system determined the locations and orientations of both the mapping and reference catheters. The location of the mapping catheter was gated to a fiducial point in the cardiac cycle and recorded relative to the location of the fixed reference catheter at that time, thus compensating for both subject and cardiac motion. For example, choosing the R wave as the fiducial point resulted in the determination of location at end diastole. In this setting, the end-diastolic location and orientation of the catheter were continuously shown on the screen of the mapping computer. By moving the catheter inside the heart, the system continuously analyzed its location and presented it to the user, thus enabling navigation without the use of fluoroscopy.
The mapping catheter was dragged over the endocardium, sequentially acquiring its tip location and its electrogram while the catheter was in stable contact with the endocardium. The LAT at each site was determined from the intracardiac electrogram. We report results that were collected from unipolar recordings filtered at 0.5 to 400 Hz. The LAT at each site was calculated as the interval between a fiducial point on the body-surface ECG or a fixed intracardiac electrogram and the steepest negative intrinsic deflection from the mapping-catheter unipolar recording. The stability of the catheter-wall contact was evaluated at every site by examination of the end-diastolic stability (the distance in millimeters between two consecutive end-diastolic locations) and the LAT stability (measured as the difference in milliseconds between two consecutive LATs). A point was added to the map only if the end-diastolic location stability was <2 mm and the LAT stability was <2 ms. The electrophysiological information was color-coded (red being the shortest LAT and purple being the longest) and superimposed on the 3D chamber geometry. The reconstruction was updated in real time with the acquisition of each new site. The map was presented as a 3D object in which interpolation of activation between adjacent sites was allowed only for nearby sites. Whenever the distance between adjacent sites was higher than a predetermined threshold, no interpolation was made, and the sites were connected by wire frames. Fig 2⇓ illustrates an example of reconstruction of the swine left ventricle.
Validation of Intracardiac Signal Recording
The intracardiac signal from the tip of the locatable catheter was correlated with a standard nonlocatable electrophysiological catheter (CardioRhythm). The two catheters were positioned as close as possible with the use of multiple x-ray projections. Both signals were simultaneously recorded, filtered, and gained identically and later were analyzed by cross-correlation methods. Correlation between signals was calculated as the cross-correlation: where C(k) is the correlation coefficient; Xi and Yi are the corresponding signals recorded from the standard and locatable catheters, respectively; X and Y are the corresponding averages of the respective signals; and n is the length of the window studied. Correlation between the two signals was determined as the peak of the resultant cross-correlation function calculated between the two signals aligned at a window of 400 ms. Altogether, 11 different sites at the left ventricle were studied.
Validation of In Vitro Location Accuracy
A special test jig was used to test the in vitro accuracy of the system and locatable-catheter capabilities (Fig 3⇓). The test jig was made of a Plexiglas block with seven different holes with known absolute locations and distances between each other. The catheter was positioned at each site, and its location was determined 25 times at each site for five different catheter rolls (protocol A). The SD and the range of the repeated measurements at each site were calculated. Next, the relative distances between each site and the six other sites were calculated as the geometric distances and compared with the known distances (protocol B).
In Vivo Validation of Location Accuracy in the Pig Studies
Studies were performed on healthy male pigs weighing 30 to 40 kg. The experimental protocol was approved by the Animal Use and Care Committee of the Technion Faculty of Medicine. All pigs were premedicated with ketamine HCl (20 mg/kg IM) and xylazine (2 mg/kg IM), anesthetized with pentobarbital (30 mg/kg), intubated, and placed on a Harvard large-animal mechanical respirator. Vascular access was obtained with vascular cutdown of the jugular veins, carotid arteries, and femoral vein and artery as needed.
At this stage, the catheter was introduced into the left ventricle. The catheter was positioned in stable contact with the endocardium, and its location was determined 20 times. This procedure was repeated at 15 sites. The standard, maximal, and average deviations at each site were calculated.
Six locatable catheters were marked with a fine-tip permanent marker along their shafts at 10-mm intervals and were used as mapping catheters. Under fluoroscopic guidance, a long 8F sheath was introduced into the RVA from the right jugular vein. The reference catheter was introduced through the left jugular vein and was positioned in the CS. The mapping catheter was introduced through the long sheath all the way down to the RVA (Fig 4⇓). At this stage, a marking on the mapping catheter was aligned with the valve of the long sheath, and the system was used to record the mapping-tip location. The mapping catheter was then withdrawn until the next mark was aligned with the valve, and the tip location was recorded again. This procedure was repeated 10 times. The distance between two sequential locations was calculated by the system as the geometric distance and was compared with the actual distance between the marks on the respective catheters. Altogether, the procedure was carried out 37 times with six different catheters and a total of 333 measured distances.
Mapping of the left ventricle. Seventy-seven ventricular activation maps were acquired, 55 during midseptal right ventricular pacing and 22 during sinus rhythm. The reference catheter was positioned, with fluoroscopic guidance, at the CS or the RVA. Using fluoroscopy, the mapping catheter was introduced into the left ventricle through the carotid artery. The first three points (two at the base, near the aortic and mitral valves, and one at the apex) were acquired with the use of fluoroscopy. No guiding catheters were used for the mapping procedure. Each activation map was reconstructed from multiple sampling points ranging from 20 to 352 points. During the pacing experiments, the catheter was navigated to the earliest site of activation without fluoroscopy. The relative location of the mapping catheter tip to the right ventricular pacing catheter was qualitatively evaluated by viewing the distance between the two catheters from multiple fluoroscopic angles. Because the two catheters were on different chambers, the distance between them depended on the thickness of the septum. The two catheters were considered to be adjacent if the maximal distance between them in the different fluoroscopic views was <12 mm (three times the length of the tip) and the mapping catheter was pointing toward the position of the pacing catheter.
Mapping the right ventricle and right atrium. A total of 11 activation maps of the right ventricle and 32 activation maps of the right atrium during sinus rhythm were acquired. The mapping catheter was introduced through the jugular or femoral veins under fluoroscopic guidance and positioned in the desired chamber. We acquired the first three points with fluoroscopy (in the right ventricle: apex, outflow tract, and tricuspid annulus; in the right atrium: near the SVC, at the right atrial junction near the IVC, and at the ostium of the CS). The rest of the mapping procedure was achieved without the use of fluoroscopy.
Results are reported as mean±SEM. Correlation between signals was calculated as the cross-correlation between two signals aligned for a window of 400 ms around the QRS complex.
Validation of Intracardiac Electrogram Recording
Using the cross-correlation method, we found a high correlation (cross-correlation coefficient=.96±.01) between the intracardiac electrograms recorded with the locatable catheter and the standard nonlocatable electrophysiological catheter.
Validation of In Vitro Location Accuracy
The in vitro studies consisted of two different protocols.
In protocol A, the dispersion of repeated location measurements at the same site was studied. As can be seen from the Table⇓, the SD of the “local cloud” of the repeated measurements ranged from 0.11 to 0.29 mm at the seven different sites, with the average SD being 0.16±0.02 mm. Maximal range extended from 0.35 to 0.89 mm, with the average maximal range being 0.55±0.07 mm.
In protocol B, the geometric distances of each site from the other six sites were calculated by the system and compared with the known distances. The mean error ranged from 0.31 to 0.71, with the total mean error being 0.42±0.05 mm.
Validation of In Vivo Location Accuracy
Similar to the in vitro measurements, we tested the reproducibility of repeated location measurements of the catheter tip while it was inside the beating heart. The SD of the local cloud was 0.74±0.13 mm during 20 consecutive location determinations at 15 different sites. The average maximal range and average mean error for all sites were 1.26±0.08 and 0.54±0.05 mm, respectively.
In this protocol, the relative distances as measured by the system while sequentially redrawing the catheter at 10-mm intervals were compared with the actual distances. The procedure was repeated several times with six different catheters. The mean error ranged from 0.63±0.07 to 0.84±0.07 mm. The overall mean error for 333 different measurements was 0.73±0.03 mm.
General information. From our initial experience, we found that after acquiring the first three points under fluoroscopic guidance, the rest of the mapping procedure could be achieved without the use of fluoroscopy. We also noted that the mapping catheter achieved stable contact with the endocardium usually after three to four heartbeats and that nonfluoroscopic navigation and mapping in the swine heart were achieved relatively fast without the use of guiding sheaths.
Results of mapping the left ventricle. Seventy-seven left ventricular activation maps were acquired, 55 during pacing of the right ventricle and septum and 22 during sinus rhythm. The geometries of all maps were similar, and all had the following characteristics: apex pointing to the right side of the chest, base located superiorly, and left atrium close to the left lateral aspect of the left ventricle (Fig 5⇓). We noted two indentations of the geometry, which corresponded anatomically to the position of the papillary muscles.
The activation sequence was similar in all maps acquired during similar conditions. During ventricular pacing, the earliest site of activation was at the site of pacing. During sinus rhythm, the earliest site of activation was on the superior part of the septum; in some pigs, a second breakthrough was noted more posteriorly on the septum. Invariably, the latest site of activation in both rhythms was on the left lateral wall close to the mitral valve annulus. The total activation time of the left ventricle varied between 40 and 80 ms during sinus rhythm and between 57 and 87 ms during paced rhythm. During pacing, when the catheter was navigated back to the site of earliest activation, its relative location was compared with the location of the right ventricular pacing electrode. Using our semiquantitative protocol, we found that in all cases its location was considered to be adjacent to the pacing electrode.
Fig 5A⇑ shows a right lateral view of a typical 3D activation map of the left ventricle during right ventricular midseptum pacing. The map was reconstructed using 60 sampling points. Note that the site of earliest activation (red) was located in the inferomedial septum adjacent to the pacing site. Activation then spread to the rest of the ventricle, with the posterolateral areas (blue and purple) being activated last. The rest of the ventricle had intermediate LATs (yellow and green).
Fig 5B⇑ represents a typical activation map of the left ventricle during sinus rhythm, reconstructed from 78 sampling points. The earliest activation site (red) was located in the anterosuperior region in all maps. Rapid conduction was noted from the site of earliest activation to the apex. The activation then spread to the rest of the endocardium, with the posterobasal wall and the AV ring being activated last (blue and purple).
Results of mapping the right atrium. The right atrial chamber geometry was similar in all cases (Fig 6⇓), showing the following details. Most of the right atrial wall area was located on the posterior aspect. The most prominent part of the anterior wall of the right atrium was the tricuspid annulus. On the posterior wall of the right atrium, there were two prominent protrusions: the CS ostium and the IVC. The SVC was mapped on the upper part of the right atrium. The earliest site of activation (SA node) was located on the posterior wall, close to the SVC connection with the right atrial wall. The activation then spread to the rest of the atrium, with relatively fast conduction along the posterior wall toward the IVC and the superior part of the right atrium toward the left atrium. The latest activation was detected at the tricuspid annulus and along the CS. Total activation time of the right atrium varied between 48 and 89 ms.
Fig 7⇓ displays three sequential steps in the propagation map of the swine right atrium, shown from a right lateral viewing angle. The red areas represent areas that have been activated. Note again the spread of the activation wave from the superolateral aspect of the atrium (SA node region) to the rest of the atrium.
Results of mapping the right ventricle. The maps of the right ventricle disclosed a similar anatomy (Fig 8⇓). All the maps had similar shapes, with the right ventricular outflow tract protruding to the left side of the pig’s chest. The base was located superiorly. Maps were acquired during sinus rhythm, and the earliest activation site was on the lower part of the septum, with fast conduction toward the apex. The septum seemed to bulge into the right ventricular chamber. The latest sites of activation were the right lateral wall and the posterobasal portion of the right ventricle.
This report presents a new method for nonfluoroscopic catheter-based endocardial mapping. The study demonstrates that the new method is highly accurate and reproducible in both in vitro and in vivo studies and that electroanatomical activation mapping of the cardiac chambers can be achieved with a high degree of precision and reproducibility without the use of fluoroscopy.
In Vitro Studies
The accuracy of location was verified by use of bench studies with a custom-made high-precision test jig (20×20×20 cm). This mapping volume was large enough to contain the human heart. First, the reproducibility of location determination at each site was determined during multiple acquisitions with different orientations and was found to be highly reproducible (SD, 0.16±0.02 mm; range, 0.55±0.07 mm). Next, the geometric distances between the locations at the different sites were calculated and compared with the actual known distances. A high degree of precision was found, with the average error being 0.42±0.05 mm.
In Vivo Studies
Correcting for Animal and Cardiac Motion
When trying to determine the mapping catheter locations, one must deal with two major motion artifacts. First, the heart is constantly beating, so the catheter location changes throughout the cardiac cycle. Second, the heart may move inside the body (respiration), and the animal itself may be moving. By gating the catheter location to a fiducial point in the cardiac cycle, we can overcome the cardiac cycle motion artifacts as long as the cycle is periodic and repeatable. To compensate for the pig’s movements and the cardiac motion within the pig, we correct the location of the mapping catheter with the location of a reference catheter placed in a fixed intracardiac site.
In Vivo Accuracy Studies
As in the in vivo studies, the reproducibility of location at a single site and the accuracy of the relative distances between different sites were tested. We found that catheter location could be determined with a high degree of precision (mean relative distance error, 0.73±0.03 mm) and good reproducibility (SD, 0.74±0.05 mm).
Significance of and Possible Reasons for Error
Several factors can influence the in vivo accuracy results. The inherent noise of the location system contributes to the overall error, together with the fact that in the heart the location of two catheters is used, thereby increasing the relative location noise. Other factors that can influence location accuracy are the reproducibility of the fiducial point on the ECG, the effect of respiration on chamber geometry, and the reproducibility of cardiac mechanics on a beat-to-beat basis. Altogether, the results presented in this study prove that this method is highly accurate and reproducible.
In addition, currently used electrophysiological catheters have 2- or 4-mm-tip electrodes. Hence, the location error reported here, which is <1 mm, is not significant, because the tip electrode actually averages electrograms from a 2- or 4-mm area.
Comparison With Other Mapping Modalities
As described earlier, current mapping methods can be divided into methods that use a large number of electrodes and acquire the electrograms simultaneously from a plurality of sites and methods that acquire electrograms sequentially.
Current techniques for mapping simultaneously from multiple sites present certain difficulties. For example, the endocardial multielectrode balloon requires open-heart surgery, heart-lung bypass, emptying of the blood from the heart, and inflation of the balloon so that the electrodes will be in direct contact with the endocardium. In addition to the invasive nature of the procedure and the risk involved, the arrhythmia often cannot be induced during surgery.
Other methods that use multielectrode arrays are the epicardial sock and the basket electrode. The spatial resolution of these methods is limited by the relatively large distances between the recording electrodes and by the fact that the absolute position of the target on the myocardium is not known once the electrodes have been removed. Furthermore, the quality of the contacts of all the electrodes with the endocardium or epicardium cannot be ensured (ie, there are sites that are not mapped). Another pitfall of these methods (especially with the basket electrode) is the relative movement between the constantly contracting heart and the electrodes. The interelectrode distances on each sidearm of the basket catheter are fixed, whereas the distances between the actual recording sites on the endocardium decrease during systole. This leads to relative movement between the recording electrode and the tissue, significantly limiting the accuracy of this mapping method.
More commonly, clinical electrophysiology is performed with catheters that are introduced percutaneously into the heart chambers. During endocardial catheter mapping of the heart, the physician sequentially positions the tip at several locations. Localization and navigation of the tip are commonly done with multiplanar fluoroscopy. The currently used method is lacking in several different aspects. Because of the need to use multiple x-ray planes, the procedure is limited as to the number of recording sites, is time-consuming, and because a single localization is made over several cardiac cycles, cannot account for beat-to-beat variability. Moreover, another major limiting factor is the inability to accurately associate (tag) a specific spatial position in the heart with its specific electrogram. We are unaware of literature quantifying the spatial resolution and accuracy of fluoroscopy-based mapping methods. A further limitation of the currently used technique is the inability to accurately position the in-dwelling catheter tip in a site that was previously mapped. Finally, the use of ionizing radiation, which in some procedures can reach high-exposure doses, is hazardous for both the patient and the physician.
The limitations of the method described in this paper are common to other methods that use sequential sampling of endocardial sites. The basic assumption is that the activation pattern and chamber geometry are constant from beat to beat. This limitation is partially addressed by the stability criteria used before the information, specifically the location and LAT stability tests, is acquired. The addition of on-line body-surface ECG cross-correlation test of consecutive QRS morphologies will increase the likelihood of proving that the propagation wave is reproducible.
The new mapping method described in this report enables, for the first time, the association of electrical and spatial endocardial information. This unique ability, together with the capability to collect a large number of sites in a relatively short period without the use of ionizing radiation, may offer several advantages for both the mapping procedure and guiding ablation treatment. First, the high spatial resolution, found in our studies to be <1 mm, enables the construction of a very detailed activation map. Second, the ability to accurately combine the 3D anatomy of the relevant chamber with its electrical activities offers unique insight into the possible role of the anatomy of the chamber and the genesis of cardiac arrhythmias. This quality may play a role in guiding anatomically based ablation procedures. Third, the ability to navigate the catheter in three dimensions without the limitations of two-dimensional fluoroscopy should reduce x-ray exposure and procedure time. Fourth, the ability to relocate accurately in three dimensions the tip of the mapping catheter with reference to the electroanatomical map that was acquired brings a unique value and advantages for ablation procedures. One may first perform the mapping procedure, next develop ablation strategies and identify the target for ablation, and finally return accurately to the desired site for the delivery of energy. This unique feature, not possible with currently used image-guided techniques, may also be used for tagging the site where the energy was delivered. Finally, the unique capability of the new method to associate relevant electrophysiological information with the appropriate spatial location in the heart may play a significant role in guiding ablation strategies, not only in focal arrhythmias but also in reentrant arrhythmias. In the latter case, pacing maneuvers rather than the site of earliest local activation guide the selection of the target for ablation. The ability to record and store a wide spectrum of relevant electrophysiological information for every sample site can become a valuable tool in the selection of the most appropriate site for ablation. Thus, relevant information such as the LAT, the unipolar and bipolar electrograms, the results of entrainment, and pace mapping can all be analyzed, stored, and associated with a specific “address” on the endocardium. This will enable selection of the best target site for ablation and then, as discussed earlier, navigating back with precision to the predefined site to deliver the therapeutic energy.
Nevertheless, although the theoretical clinical advantages of the new method for mapping and ablations are obvious, future prospective studies will have to prove its clinical utility. This study opens the way to investigate the potential impact of the new method on specific clinical parameters such as reduction of fluoroscopy and procedure times, the reduction of number of radiofrequency applications, and success rate.
In conclusion, the method described in this paper combines, for the first time, electrophysiological information with the 3D anatomy of the heart chambers. This will allow better understanding of the mechanisms involved in the genesis of arrhythmias and the influence of pharmacological and nonpharmacological therapies.
Selected Abbreviations and Acronyms
|IVC||=||inferior vena cava|
|LAT||=||local activation time|
|RVA||=||right ventricular apex|
|SVC||=||superior vena cava|
This work was supported by a grant from Biosense.
- Received September 23, 1996.
- Accepted November 18, 1996.
- Copyright © 1997 by American Heart Association
Lewis T, Rothschild MA. The excitatory process in the dog’s heart, II: the ventricles. Philos Trans R Soc Lond B Biol Sci. 1915;206:181-226.
Josephson ME. Use of electrophysiological testing to select antiarrhythmic drug therapy for ventricular arrhythmia. In: Rosen MR, Janse MJ, Wit AL, eds. Cardiac Electrophysiology: A Textbook. Mount Kisco, NY: Futura Publishing Co; 1990:1137-1158.
Jackman WM, Wang X, Friday KJ, Roman CA, Moulton KP, Beckman KJ, McClelland JH, Twidale N, Hazlitt HA, Prior MI, Margolis PD, Calame JD, Overholt ED, Lazzara R. Catheter ablation of accessory atrioventricular pathways (Wolff-Parkinson-White syndrome) by radiofrequency current. N Engl J Med. 1991;324:1605-1611.
Gallagher JJ, Svenson RH, Kasell JH, German LD, Bardy GH, Broughton A, Critelli G. Catheter technique for closed-chest ablation of the atrioventricular conduction system: a therapeutic alternative for the treatment of refractory supraventricular tachycardia. N Engl J Med. 1982;306:194-200.
Flowers NC, Horan LG. Body surface potential mapping. In: Zipes DP, Jalife J, eds. Cardiac Electrophysiology: From Cell to Bedside. Philadelphia, Pa: WB Saunders; 1995:1049-1067.
Durrer D, van Dam RT, Freud GE, Janse MJ, Meijler FL, Arzbaecher RC. Total excitation of the isolated human heart. Circulation. 1970;41:899-912.
Cox JL, Canavan TE, Schuessler RB, Cain ME, Lindsay BD, Stone C, Smith PK, Corr PB, Boineau JP. The surgical treatment of atrial fibrillation, II: intraoperative electrophysiologic mapping and description of the electrophysiologic basis of atrial flutter and atrial fibrillation. J Thorac Cardiovasc Surg. 1991;101:406-426.
Gallagher JJ, Kasell JH, Cox JL, Smith WM, Ideker RE, Smith WM. Techniques of intraoperative electrophysiologic mapping. Am J Cardiol. 1982;49:221-240.
Hafala R, Sarvard P, Tremblat G, Page P, Cardinal R, Molin F, Kus T, Nadean R. Three distinct patterns of ventricular activation in infarcted human hearts: an intraoperative cardiac mapping study during sinus rhythm. Circulation. 1995;91:1480-1494.
Konings KTS, Kirchhof CJHJ, Smeets JRLM, Wellens HJJ, Penn OC, Allessie MA. High-density mapping of electrically induced atrial fibrillation in humans. Circulation. 1994;89:1665-1680.