|Year : 2019 | Volume
| Issue : 2 | Page : 48-54
The advanced reference annotation algorithm: A novel approach to reference annotation for electroanatomic mapping
Martin Aguilar1, Jonathan Yarnitsky2, Lior Botzer2, Vladimir Rubinstein2, Elad Nakar2, Vias Markides3, Nicolas Derval4, Matteo Anselmino5, Jeffrey Winterfield6, Daniel Melby7, Ibrahim Marai8, Mahmoud Suleiman9, Christian Meyer10, Paul C Zei1
1 Department of Medicine, Cardiac Arrhythmia Service, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA
2 Biosense Webster, Inc., Haifa, Israel
3 Department of Medicine, Royal Brompton Hospital, London, UK
4 Department of Medicine, Hopital Cardiologique du Haut Lévêque, CHU Bordeaux, Pessac, France
5 Division of Cardiology, Department of Medical Sciences, University of Turin, Turin, Italy
6 Department of Medicine, Medical University of South Carolina, Charleston, SC, USA
7 Department of Medicine, Minneapolis Heart Institute, Minneapolis, MN, USA
8 Department of Medicine, Baruch Padeh Medical Center, Poriya, Israel
9 Department of Medicine, Rambam Medical Center, Haifa, Israel
10 Department of Cardiology-Electrophysiology, University Heart Center, University Hospital Hamburg-Eppendorf, Berlin, Germany; Deutsches Zentrum Für Herz-Kreislauf-Forschung (German Center for Cardiovascular Research), Partner Site Hamburg/Kiel/Lübeck, Germany
|Date of Submission||19-Feb-2020|
|Date of Acceptance||01-Apr-2020|
|Date of Web Publication||15-Jun-2020|
Dr. Paul C Zei
Cardiac Arrhythmia Service, Brigham and Women's Hospital, Harvard Medical School, 70 Francis Street, Boston, MA 02115
Source of Support: None, Conflict of Interest: None
Background: Reliable reference annotation is critical for accurate activation mapping. Currently used referencing algorithms can be limited by suboptimal detection and stability performance. The advanced reference annotation (ARA) algorithm, a novel algorithm using a weighted reference across multiple electrodes, has been developed to optimize reference annotation. Materials and Methods: To evaluate ARA, recordings using CARTO from 26 clinical cases with complex cardiac arrhythmias, representing mapping of various rhythms, were segmented into test vectors consisting of roughly 62,000 annotation events. These were annotated by an expert clinician (gold standard [GS]) and compared with the legacy/ARA algorithms on detection rate and stability and positive predictive value (PPV). Results: The ARA algorithm detection rate uniformly outperformed legacy, when compared with GS (97 ± 4% vs. 81 ± 19%, respectively; P = 0.001). ARA was performed with high fidelity with an average stability metric (the percentage of true positive ARA annotations within 10 ms of the GS annotation) of 98 ± 3% with most test vectors achieving perfect (100%) stability. Overall, the PPV of ARA annotations was 98 ± 4%; nearly all ARA-annotated activation corresponded to clinically observed events; ARA was superior to legacy across all analyzed test vectors (98 ± 4% vs. 88 ± 23%, P = 0.004); all ARA test vector groups had PPV >90%. Conclusion: The ARA algorithm outperformed the clinical standard, compared to an expert clinician GS. These improvements may translate into greater mapping accuracy/efficiency and procedural outcomes in diagnosis of complex cardiac arrhythmias.
Keywords: Activation mapping, catheter ablation, electroanatomic mapping, reference annotation algorithm
|How to cite this article:|
Aguilar M, Yarnitsky J, Botzer L, Rubinstein V, Nakar E, Markides V, Derval N, Anselmino M, Winterfield J, Melby D, Marai I, Suleiman M, Meyer C, Zei PC. The advanced reference annotation algorithm: A novel approach to reference annotation for electroanatomic mapping. Int J Heart Rhythm 2019;4:48-54
|How to cite this URL:|
Aguilar M, Yarnitsky J, Botzer L, Rubinstein V, Nakar E, Markides V, Derval N, Anselmino M, Winterfield J, Melby D, Marai I, Suleiman M, Meyer C, Zei PC. The advanced reference annotation algorithm: A novel approach to reference annotation for electroanatomic mapping. Int J Heart Rhythm [serial online] 2019 [cited 2020 Oct 26];4:48-54. Available from: https://www.ijhronline.org/text.asp?2019/4/2/48/286762
| Introduction|| |
Electroanatomic mapping has become an essential component of the diagnostic and therapeutic approach to complex cardiac arrhythmias., Activation mapping is used to determine the site of earliest activation (focal tachycardias and ectopic rhythms), to identify the critical isthmus (reentrant arrhythmias), or more generally to identify arrhythmic circuits., Timing of events during mapping is relative to a fixed fiduciary reference against which signals of interest are timed., Temporal and anatomic stability, resulting in reproducible detection of the reference annotation, is of paramount importance in creating reliable activation maps, as variability in the reference will be propagated to the timing of all signals.
Most currently used referencing algorithms search for specific features within a signal (i.e., local minimum/maximum or minimum/maximum dV/dt) to annotate the reference on a beat-by-beat fashion. Unfortunately, these algorithms are sensitive to changes in signal morphology and artefactual signals (i.e., pacing) and underperform when referencing complex electrograms or overlapping atrial and ventricular signals. For example, a reference signal with two local maxima can lead to substantial error if the referencing algorithm selects different peaks on different beats (or incorrectly times the ventricular for the atrial activation). Moreover, because of inherent limitations in the currently used algorithms, they fail to annotate a significant number of tachycardia cycles, even for simple arrhythmias such as typical atrial flutter, reducing mapping efficiency, both through lack of annotation and consequent variation in perceived cycle length stability. This is particularly relevant when tachycardias are nonsustained, limiting mapping time. Although current algorithms generally perform quite well, ultimately, these limitations may negatively impact procedural outcomes.
The advanced reference annotation (ARA) algorithm was developed to overcome the shortcomings of available annotation algorithms and provide more stable and consistent reference annotation with a high detection rate. The ultimate goal of the ARA algorithm is to provide higher fidelity activation maps to facilitate the diagnosis and catheter-based treatment of cardiac arrhythmias.
The ARA algorithm is a novel approach to reference calculation aimed at increasing activation detection and annotation stability; it is used for reference annotation of both body surface and intracardiac signals when mapping with CARTO3 V7 system (Biosense Webster, Irvine, CA). The algorithm evaluates signal quality on a beat-by-beat basis to determine the channels with optimal amplitude and signal-to-noise ratio. Multiple selected channels are used to produce an integrative signal for each timing event, the activity signal, segmenting and emphasizing regions of activity. In brief, the activity signal is calculated as the sum (ventricular signal) or product (atrial signal) of the absolute values of dV/dt of the selected input signals [Figure 1]a. For ventricular signals, the precordial electrocardiogram surface leads (V1–V6) are used as input to the ARA algorithm. Given that ventricular activation on the surface electrocardiogram is typical fairly synchronous, the sum of the absolute values of dV/dt provides excellent discrimination. For atrial signals, both precordial leads and intracardiac inputs are used. The precordial signals are used to separate the ventricular from the atrial activation on the intracardiac signals. Given that atrial activation along the coronary sinus catheter is temporally dispersed, the product of the absolute values of dV/dt provides superior discrimination than the sum [Figure 1]b. This is because the sum of the temporally disperse coronary sinus signals would result in a wide and low-amplitude signal with ambiguous reference timing. This calculation method stands to gain a sharp activity signal which is then utilized to identify atrial segments of activation. Finally, thresholds dynamically measure activity and noise levels to maximize true detection and minimize erroneous detections. The activation annotation is timed to the “center of energy” of the activity signal.
|Figure 1: Conceptual description of the advanced reference annotation algorithm. (a) Ventricular-based reference annotation with the advanced reference annotation algorithm. Ventricular channels with good signal-to-noise ratio enter the calculation (left). The sum (or product) of dV/dt is used to generate the activity signal (center) and define the timing of activation based on the center of energy of the activity signal (right). (b) Atrial-based reference annotation with the advanced reference annotation-algorithm. The ventricular activation is determined as per the ventricular-based referencing algorithm. The product of dV/dt is used to generate the activity signal and define the timing of activation|
Click here to view
On initialization, the ARA algorithm performs a preliminary signal analysis to differentiate true activation from noise such that only channels demonstrating a predominant activation are annotated based on their own center of energy. Only channels showing close temporal activations are grouped to form the primary annotation by averaging the single annotations of that group. Following the calculation of the primary annotation, for all subsequent activations, the dynamically selected set of predominant channels is annotated based on the channels' center of energy. Subsequent reference annotations are based on the average cycle length from the last similar group of channels.
The purpose of this study was to test the performance of the ARA algorithm in a wide range of commonly encountered mapping scenarios and compare it to a common clinically used algorithm (legacy algorithm) and expert clinician annotations (gold standard [GS]) in the diagnosis of cardiac arrhythmias.
| Materials and Methods|| |
Clinical cases used for annotation validation
Intracardiac and/or body surface recordings from 26 clinical cases grouped into 17 categories were collected for analysis [Table 1]. Cases were segmented into 94 5-min test vectors during the mapping phase of each study. Test vectors were selected to represent various clinically relevant rhythms including different arrhythmias, cycle lengths, signal morphologies, pacing events, and interferences (such as defibrillation). For each test vector, the beat-to-beat activation reference was manually annotated by an expert clinician using a custom annotation tool (CARTO custom tool, Biosense Webster, Irvine, CA, USA), marking based on the maximum or minimum value of the activation for a selected channel of their choice. In other words, the expert clinician visually annotated the reference timing of each beat on the basis of prominent and reproducible features of the input signal; these annotations were marked using the custom annotation tool as the GS. Overall, a total of 61,608 atrial annotations were marked on intracardiac signals and 7468 ventricular annotations marked on the precordial leads. The same test vectors were then annotated by the currently clinically used legacy algorithm and the ARA algorithm; the annotations were then compared. We used three metrics to quantify the performance of the ARA algorithm: (i) detection rate, (ii) detection stability, and (iii) positive predictive value (PPV). This was a retrospective study using a deidentified data set. Institutional board review and patient consent were not required.
|Table 1: Twenty-six cases were segmented into 945-minute test vectors categorized into 17 groups|
Click here to view
The annotation detection metric
The detection rate metric was used to assess the performance of the annotation algorithm to detect complex activation, irrespective of the position of the annotation. An annotation was categorized as a true positive (TP) annotation if it fell within 150 ms of an intracardiac or 100 ms of a body surface GS (expert clinician) annotation. If more than one legacy/ARA annotation fell within 150/100 ms of a GS annotation, the closest legacy/ARA annotation was marked as a TP and the other considered false positive (FP); legacy/ARA annotations not marked as either TP or FP were labeled as false negative (FN) annotations. The legacy/ARA detection rate was quantified as their respective sensitivity (Sn = TP/[TP + FN]); the higher the detection rate (sensitive), the better for this metric.
The annotation stability metric
The stability analysis aimed to evaluate the ability of the annotation algorithm to maintain a consistent temporal annotation position for a given rhythm versus the GS (expert clinician) annotation. Initially, all activations detected per test vector were separated into groups of similar morphology based on mathematical correlation product, so that each group represented a specific rhythm or form of activation. The stability metric was defined as the percentage of TP ARA annotations falling within 10 ms of the GS annotation. Given that the annotation method by the expert clinician and the ARA algorithm is based on either maximum/minimum of the signal or “center of energy,” respectively, the constant average offset between the ARA annotations and the GS per group was subtracted, as this does not affect annotation stability or the resulting activation map.
The annotation positive detection rate metric
The positive detection rate was used to quantify the ability of the algorithm to correctly annotate points (TPs) without contaminating with incorrect annotations (FPs). The positive detection metric was defined as the PPV (PPV = TP/[TP + FP]) for each 5-minute test vector.
The following patents were applied to ARA algorithm: (1) accurate time annotation of intracardiac ECG signals (patent number: 8700136) and (2) determination of reference annotation time from multi-channel electrocardiogram signals (patent number: 9259165).
The data were presented as the mean ± standard deviation. The analyses were conducted using GraphPad Prism 8 (GraphPad, San Diego, CA, USA).
| Results|| |
The detection rate quantifies the ability of the algorithm to detect complex activation as compared to the GS, regardless of the timing of the annotation. The ARA algorithm detection rate across the set of test vectors was excellent with 97 ± 4% detection versus the expert clinician's annotation. The ARA algorithm's overall detection rate outperformed the legacy algorithm (97 ± 4% vs. 81 ± 19%, respectively; P = 0.001). Focusing on the detection performance for specific rhythms, the ARA algorithm was uniformly superior to the legacy algorithm and maintained excellent detection rates across the range of intrinsic and paced rhythms under consideration [Figure 2]a. The lowest detection rate was with test vectors from a supraventricular rhythm [Figure 2]a; group 5], but was nevertheless satisfactory (92 ± 6%) and outperformed the legacy algorithm (62 ± 9%). [Figure 3]a shows an example comparing the expert clinician annotations (dotted yellow) to successfully detected annotation with the legacy (brown) and ARA (green) algorithms during typical atrial flutter. [Figure 3]b shows an example where the ARA algorithm fails to annotate the reference on certain beats and annotates with significant interbeat variations during atrial tachycardia.
|Figure 2: Comparison of detection performance between the legacy and advanced reference annotation algorithm. (a) Detection rate of activation signals for the legacy (red) versus advanced reference annotation (blue) algorithm for the different test groups. advanced reference annotation outperformed the legacy algorithm detection rate across all groups (97 ± 4% vs. 81 ± 19%; P = 0.001). (b) Stability metric for the advanced reference annotation algorithm for the different test groups. Advanced reference annotation outperformed the legacy algorithm detection rate across all groups (98 ± 3%). (c) Positive detection rate for the legacy (red) versus advanced reference annotation (blue) algorithm for the different test groups. The advanced reference annotation algorithm outperformed the legacy algorithm detection rate across all groups (98 ± 4% vs. 88 ± 23%; P = 0.001). Twenty-six clinical cases grouped into 17 categories are shown in Table 1. Groups 12–17 represent clinical test vectors that were manipulated to challenge the advanced reference annotation algorithm (by adding noise, adding segments of catheter disconnection, for example) and were therefore not analyzed using the legacy algorithm (grey bars)|
Click here to view
|Figure 3: Representative examples of legacy versus advanced reference annotation annotations. (a) Example of correct and stable detection during typical atrial flutter coronary sinus-based-mapping; gold standard (yellow dotted), legacy algorithm (brown), advanced reference annotation algorithm (green). (b) Example of unstable and missed annotations by the advanced reference annotation algorithm (green) versus the gold standard (dotted yellow) during CS-based-mapping of an atrial tachycardia|
Click here to view
We then compared the ARA algorithm annotations' stability metric, referring to the algorithm's ability to provide a consistent position of the annotation relative to the activation morphology, for a given rhythm. Overall, the ARA algorithm was performed with very high fidelity with an average stability metric of 98 ± 3% [Figure 2]b. All test vector groups except for one had stability >95% with several achieving perfect (100%) stability. The test vector group with the lowest stability metric was with intermittent ventricular pacing at 91 ± 9%.
Positive detection rate metric
An improvement in detection rate is only meaningful if the annotated activations correspond to the actual clinically observed activation. The positive detection rate was quantified using the positive predicted value (PPV) of ARA algorithm versus the GS. Overall, the PPV of ARA annotations was 98 ± 4%, meaning that nearly all the ARA annotated activation corresponded to clinically observed events; the ARA algorithm was superior to the legacy algorithm across all analyzed test vectors [98 ± 4% vs. 88 ± 23%, P = 0.004; [Figure 2]c. All test vector groups had PPV >90% with the lowest PPV observed for alternating atrial rhythm (group 17; 91 ± 5%), whereas the lowest PPV with the legacy algorithm was much lower (64 ± 34%).
The superior detection rates observed with the ARA algorithm translated into higher point density maps versus the legacy algorithm for a given study. [Figure 4] shows a representative example comparing the activation maps during atypical atrial flutter generated with legacy annotations and ARA annotations. The legacy-generated map contains 2960 points, whereas the ARA-generated map consists of 4906 points (65% increment). The ARA-generated map is qualitatively similar to the legacy-generate map but with significantly higher resolution for the same mapping time.
|Figure 4: Activation map for an atypical atrial flutter obtained using the legacy annotation algorithm (a) and advanced reference annotation (b). The legacy-generated map contains 2960 points whereas the advanced reference annotation-generated map consists of 4906 points (65% increment). The maps are qualitatively similar with the advanced reference annotation-generated map having significantly higher resolution for the same mapping time|
Click here to view
| Discussion|| |
The main finding of this study is that the ARA algorithm provides superior overall annotation detection rate, reference stability, and positive detection rate when compared to a currently used annotation algorithm (legacy). Moreover, ARA-based annotation closely approximates GS annotation as determined by expert clinicians with detection rates well in excess of 90% (average 97%) across the analyzed test vector groups. The ARA algorithm stability and positive detection rate performed exceeding well across a wide range of clinically relevant mapping scenarios such as typical and atypical atrial flutter, atrial tachycardia with variable atrioventricular conduction, ventricular premature contractions and tachycardia, as well as during pacing during maneuvers.
Activation mapping is an integral component of the catheter-based diagnostic and therapeutic approach to both focal and reentrant arrhythmias.,,, Paramount to the diagnostic accuracy of activation mapping is having a consistent and stable reference relative to which events are timed.,,,,,,,, However, most currently used algorithms (i.e., legacy algorithm for the CARTO system) use single-channel signal characteristics, such as the local maximum or minimum, to annotate events leading potentially to failure to detect the fiduciary point, erroneous detection of an event as the reference and/or sub-optimal beat-to-beat stability in timing of the fiduciary point., Although the current reference algorithms perform well, these errors may result in increased mapping time, inaccurate maps, incorrect arrhythmia mechanism or circuit delineation, and, ultimately, unsuccessful ablation.,,, The ARA algorithm was developed in response to these shortcomings to improve the detection rate, stability, and consistency of reference annotations and ultimately enhance activation map fidelity, providing significant improvement beyond previous algorithms.
The legacy algorithm, the annotation algorithm currently used by CARTO, utilizes a single channel to identify a prespecified feature of the signal such as the local maximal/minimum or maximal positive/negative dV/dt. After detection of an event, an exponentially decreasing detection threshold is initiated until the next activation and the process is repeated in an iterative fashion. This algorithm's detection performance meets current clinical standards. Nevertheless, the legacy algorithm can fail in commonly encountered clinical scenarios, leading to failure to detect a signal or incorrect timing of the annotation. For example, during pacing maneuvers, the detection threshold is initiated at a supraphysiological value potentially leading to undersensing of physiological signals during the decay of the detection threshold. Similarly, commonly encountered arrhythmias such as typical atrial flutter can challenge the legacy algorithm because the ventricular far-field may cause failure to detect every other atrial electrogram (in the case of 2:1 atrioventricular conduction). Finally, when using a single-channel input, as is the case with the legacy algorithm, the reliability of the annotations is highly dependent on changes in the signal morphology, noise level, and reference catheter physical stability, which are ubiquitous concerns during clinical mapping.
The ARA algorithm introduces fundamentally novel features to optimize signal detection and annotation stability. Most notable is the utilization of the additive value of inputting multiple channels into the algorithm (vs. single channel) to construct a much more robust and reliable reference annotation. Moreover, the algorithm dynamically changes the channels used for analysis on a beat-by-beat basis to input the signals with the highest discriminatory value. The method used to identify the activation is fundamentally different than with currently used algorithms. Instead of looking for a specific characteristic of the input signal (i.e., local maximal/minimum or maximal positive/negative dV/dt), the ARA algorithm analyzes the derivatives of the multichannel input to define a “center of energy” timing for the reference annotation. This method may be intuitively more robust than finding a specific feature in a physiological, intrinsically variable single-channel input because of the additional information provided by multichannel analysis and the mathematically generic (i.e., insensitive to small perturbations in signal morphology) way of defining the reference. The results of the present study provide empiric validation for this novel algorithm over a wide range of commonly encountered mapping scenarios.
Interestingly, the very nature of the ARA algorithm makes it such that the reference annotations do not correspond to any specific signal/event, as opposed to the legacy-based annotations. However, the ARA approach is valid because the absolute timing of the reference in the cardiac cycle is irrelevant for activation mapping. In other words, contrary to voltage mapping where the electrogram characteristic contains diagnostically significant information, there is no advantage in timing the reference on any specific electrogram during activation mapping. Rather absolute beat-to-beat stability is the most valuable property of the reference annotation. The concept of “center of energy” is analogous to that of center of mass (or gravity) which is used in classical mechanics to represent a complex object as a point of equal mass for the purposes of calculating the effects of external forces applied to that object. In the context of the present study, the “center of energy” is a conceptually similar construct, allowing one to define a temporal reference of a multi-input signal.
For the purposes of validation of the ARA algorithm, we employed expert clinician annotations of the test vectors as our GS. Interestingly, in cases where the reference signal is complex, it is conceivable that the ARA algorithm annotations may be more stable that those marked by a clinician, given that the clinician's annotation will usually be based on a single-channel signal characteristic and the ARA algorithm incorporates the information from multiple channels. Moreover, the ARA algorithm does not add a significant computational burden and should therefore not impact the performance of the mapping software. Importantly, the ARA algorithm is not currently optimized to map during atrial fibrillation or accessory pathways with intracardiac reference, atrioventricular nodal reentrant tachycardia with intracardiac reference, or ventricular rhythms with intracardiac reference. This is related to the challenge that a reference algorithm may have to discriminate atrial and ventricular activation on the intracardiac signals in those tachycardias, leading to potentially somewhat inferior performance.
The clinical significance of these findings is that the ARA algorithm provides superior annotation detection rate and stability under a wide range of clinical scenarios improving activation mapping efficiency and fidelity.
In utilizing expert physician adjudication of reference signal annotation, the utilized GS is assumed to have the highest degree of accuracy among the annotation methodologies. It may be possible that physician-based annotations may have intrinsic inaccuracies, as typically a single reference channel electrogram was used to determine reference timing, and human error may be introduced. As a result, the comparison of both the legacy and ARA annotations may not be entirely representative. However, it is not likely that a more accurate GS can be employed.
As with all currently used annotation methodologies, significant perturbations in the tachycardia, including ectopic beats during tachycardia, changes in cycle length, and shifts to alternative tachycardias, are not automatically detected. A primary goal of ARA is to increase the sensitivity and specificity of identifying true activation events, whether through intracardiac or surface signal detection, and as a result, potentially improving electroanatomic map accuracy and mapping time. As such, ARA will not completely eliminate the ultimate need for physician input to discriminate outlier beats. Other existing algorithms in the CARTO mapping system (ConfiDENSE, for instance) can address these issues to a large extent.
While clearly demonstrating the superiority of ARA over legacy annotation algorithms, the nature of the study, based on retrospective analysis, means that improvements in mapping time, map accuracy, and most importantly ablation success and clinical outcomes cannot be demonstrated. Such endpoints are in fact quite difficult to evaluate in an alternative, prospective setting, and therefore they cannot be assessed.
| Conclusion|| |
In this study, the ARA, a novel reference annotation methodology for activation mapping of arrhythmia, was evaluated using arrhythmias recorded during clinical cases. In comparison to the currently utilized standard annotation system (legacy), ARA outperformed legacy in several measures, including measures of sensitivity to detection of true reference annotation events, as well as measures of stability of the reference annotation, when compared to a GS of expert physician manual annotation. These findings were evident across a spectrum of arrhythmias. These findings suggest that such an annotation algorithm may provide improvements in electroanatomic map efficiency, map accuracy, efficiency in electroanatomical map generation, and potentially clinical procedural outcomes. Further evaluation on the impact of the ARA algorithm on procedural endpoints is warranted.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
Institutional review board statement
This was a retrospective study using a de-identified data set, thus institutional board review was not required.
Declaration of patient consent
The need for written informed consent was waived owing to the retrospective nature of the study.
| References|| |
Kistler PM, Rajappan K, Harris S, Earley MJ, Richmond L, Sporton SC, et al
. The impact of image integration on catheter ablation of atrial fibrillation using electroanatomic mapping: A prospective randomized study. Eur Heart J 2008;29:3029-36.
Del Carpio Munoz F, Buescher TL, Asirvatham SJ. Three-dimensional mapping of cardiac arrhythmias: What do the colors really mean? Circ Arrhythm Electrophysiol 2010;3:e6-11.
Bhakta D, Miller JM. Principles of electroanatomic mapping. Indian Pacing Electrophysiol J 2008;8:32-50.
Dong J, Calkins H, Solomon SB, Lai S, Dalal D, Lardo AC, et al
. Integrated electroanatomic mapping with three-dimensional computed tomographic images for real-time guided ablations. Circulation 2006;113:186-94.
Nademanee K, McKenzie J, Kosar E, Schwab M, Sunsaneewitayakul B, Vasavakul T, et al
. A new approach for catheter ablation of atrial fibrillation: Mapping of the electrophysiologic substrate. J Am Coll Cardiol 2004;43:2044-53.
Wittkampf FH. Image integration in 3D catheter mapping systems: Proof of the pudding. Eur Heart J 2008;29:2957-8.
de Groot NM, Schalij MJ, Zeppenfeld K, Blom NA, Van der Velde ET, Van der Wall EE. Voltage and activation mapping: How the recording technique affects the outcome of catheter ablation procedures in patients with congenital heart disease. Circulation 2003;108:2099-106.
de Groot NM, Bootsma M, van der Velde ET, Schalij MJ. Three-dimensional catheter positioning during radiofrequency ablation in patients:First application of a real-time position management system. J Cardiovasc Electrophysiol 2000;11:1183-92.
Dorostkar PC, Cheng J, Scheinman MM. Electroanatomical mapping and ablation of the substrate supporting intraatrial reentrant tachycardia after palliation for complex congenital heart disease. Pacing Clin Electrophysiol 1998;21:1810-9.
Nakagawa H, Jackman WM. Use of a three-dimensional, nonfluoroscopic mapping system for catheter ablation of typical atrial flutter. Pacing Clin Electrophysiol 1998;21:1279-86.
Reithmann C, Hoffmann E, Dorwarth U, Remp T, Steinbeck G. Electroanatomical mapping for visualization of atrial activation in patients with incisional atrial tachycardias. Eur Heart J 2001;22:237-46.
Shah DC, Jaïs P, Haïssaguerre M, Chouairi S, Takahashi A, Hocini M, et al
. Three-dimensional mapping of the common atrial flutter circuit in the right atrium. Circulation 1997;96:3904-12.
Smeets JL, Ben-Haim SA, Rodriguez LM, Timmermans C, Wellens HJ. New method for nonfluoroscopic endocardial mapping in humans: Accuracy assessment and first clinical results. Circulation 1998;97:2426-32.
Stevenson WG, Delacretaz E, Friedman PL, Ellison KE. Identification and ablation of macroreentrant ventricular tachycardia with the CARTO electroanatomical mapping system. Pacing Clin Electrophysiol 1998;21:1448-56.
Weiss C, Willems S, Rueppel R, Hoffmann M, Meinertz T. Electroanatomical Mapping (CARTO) of ectopic atrial tachycardia: Impact of bipolar and unipolar local electrogram annotation for localization the focal origin. J Interv Card Electrophysiol 2001;5:101-7.
Namboodiri N. Electroanatomic contact mapping: How to use optimally to recognise the arrhythmia mechanism? Indian Pacing Electrophysiol J 2010;10:1-7.
Cantwell CD, Roney CH, Ng FS, Siggers JH, Sherwin SJ, Peters NS. Techniques for automated local activation time annotation and conduction velocity estimation in cardiac mapping. Comput Biol Med 2015;65:229-42.
Katritsis G, Luther V, Kanagaratnam P, Linton NW. Arrhythmia Mechanisms Revealed by Ripple Mapping. Arrhythm Electrophysiol Rev 2018;7:261-4.
Nakagawa H, Shah N, Matsudaira K, Overholt E, Chandrasekaran K, Beckman KJ, et al
. Characterization of reentrant circuit in macroreentrant right atrial tachycardia after surgical repair of congenital heart disease: Isolated channels between scars allow “focal” ablation. Circulation 2001;103:699-709.
Corrado D, Basso C, Leoni L, Tokajuk B, Bauce B, Frigo G, et al
. Three-dimensional electroanatomic voltage mapping increases accuracy of diagnosing arrhythmogenic right ventricular cardiomyopathy/dysplasia. Circulation 2005;111:3042-50.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]