Atrial Fibrillation Electrograms Research Articles

Fractional-order modeling of myocardium structure effects on atrial fibrillation electrograms

Atrial fibrillation (AF) is the most common cardiac arrhythmia with mechanisms of initiation and sustaining that are not fully understood. The clinical procedure for AF contemplates the analysis of the atrial electrograms, whose morphology has been correlated with the underlying structure of the atrial myocardium. This study employs a mathematical model incorporating fractional calculus to simulate cardiac electrical conduction, accounting for tissue structural inhomogeneities using complex-valued orders. Simulations of different wavefront propagation patterns were performed, and virtual electrograms were analyzed using an asymmetry factor. Our results evinced that the shapes of the action potential and the propagating wavefront can be modulated through the fractional order under both healthy and AF conditions. Moreover, the asymmetry factor changes with variations in the fractional order. For a given propagation pattern under AF conditions, variation intervals for the asymmetry factor can be generated by forming sets of simulations with different configurations for the fractional order, representing diverse samples of atrial tissue with varying degrees of structural heterogeneity. This approach successfully reproduces the electrogram negative deflection predominance seen in AF patients, which standard integer-order models cannot predict. Our fractional-order conduction model aligns with the effects of atrial structure on the electrical dynamics observed in clinical AF. Therefore, it offers a valuable tool for studying cardiac electrophysiology, encompassing both electrical and structural interactions of the tissue within a unified model.

Stochastic time series analysis of atrial fibrillatory electrograms in paroxysmal atrial fibrillation

Abstract Background/Introduction Regional dominant frequency and instantaneous frequency modulation of short fractionated atrial electrograms have been associated with ablation targets and the prognosis of atrial fibrillation (AF). However, inconsistent and not fully comprehensible results have been reported. Purpose To construct a simple computational algorithm for the analysis of continuous fibrillatory atrial electrograms and evaluate it in clinical practice, among patients with paroxysmal AF. Methods Before the first cryo-balloon ablation, 3-minute continuous local electrogram recordings during pace-induced AF were obtained by the circumferential catheter from the left atrial appendage (LAA) and each pulmonary vein: left superior (LSPV), left inferior (LIPV), right superior (RSPV) and right inferior (RIPV). Following time- and frequency-domain analysis, non-linear detrended fluctuation analysis (DFA α) with analysis for short-term (DFA α1) and long-term fluctuations (DFA α2) was performed. The signals were detrended by subtracting time series polynomial fit and badpass filtered in the MATLAB environment (cut-off voltage 0.1 mV to avoid scar and artifacts). Results Fifteen patients were studied. Among different atrial areas, higher mean peak rate (pr) along with lower standard deviation of between peaks intervals (sdpp) and root square of peak-to-peak differences (rmppd) were observed in the LAA and left veins (Fig. 1). In patients with longer AF history (> 2 yrs), lower α, α1 and α2 exponents of DFA were observed (Fig. 2). Conclusions This algorithm depicted dissimilar impact on atrial electrical activity between different atrial areas. Patients with longer AF history had less self-similarity in local LAA electrograms. This might indicate possible ablation targets and reveal patients with greater AF burden and worse arrhythmic prognosis.Leftheriotis - Figure 1Leftheriotis - Figure 2

Use of Atrial Fibrillation Electrograms and T1/T2 Magnetic Resonance Imaging to Define the Progressive Nature of Molecular and Structural Remodeling: A New Paradigm Underlying the Emergence of Persistent Atrial Fibrillation.

The temporal progression states of the molecular and structural substrate in atrial fibrillation (AF) are not well understood. We hypothesized that these can be detected by AF electrograms and magnetic resonance imaging parametric mapping. AF was induced in 43 dogs (25-35 kg, ≥1 year) by rapid atrial pacing (RAP) (3-33 weeks, 600 beats/min), and 4 controls were used. We performed high-resolution epicardial mapping (UnEmap, 6 atrial regions, both atria, 130 electrodes, distance 2.5 mm) and analyzed electrogram cycle length, dominant frequency, organization index, and peak-to-peak bipolar voltage. Implantable telemetry recordings were used to quantify parasympathetic nerve activity over RAP time. Magnetic resonance imaging native T1, postcontrast T1, T2 mapping, and extracellular volume fraction were assessed (1.5T, Siemens) at baseline and AF. In explanted atrial tissue, DNA oxidative damage (8-hydroxy-2'-deoxyguanosine staining) and percentage of fibrofatty tissue were quantified. Cycle length and organization index decreased (R=0.5, P<0.05; and R=0.5, P<0.05; respectively), and dominant frequency increased (R=0.3, P n.s.) until 80 days of RAP but not thereafter. In contrast, voltage continued to decrease throughout the duration of RAP (R=0.6, P<0.05). Parasympathetic nerve activity increased following RAP and plateaued at 80 days. Magnetic resonance imaging native T1 and T2 times increased with RAP days (R=0.5, P<0.05; R=0.6, P<0.05) in the posterior left atrium throughout RAP. Increased RAP days correlated with increasing 8-hydroxy-2'-deoxyguanosine levels and with fibrosis percentage (R=0.5, P<0.05 for both). A combination of AF electrogram characteristics and T1/T2 magnetic resonance imaging can detect early-stage AF remodeling (autonomic remodeling, oxidative stress) and advanced AF remodeling due to oxidative stress and fibrosis.

Abstract 17206: Targeted Ablation at Regions of High Oxidative Injury in Persistent Atrial Fibrillation in a Canine Model — A New Electrogram Based Therapeutic Strategy in Atrial Fibrillation?

Introduction: Oxidative stress (OS) plays an important role in the development of atrial fibrillation (AF). However, it is currently not possible to detect regions of high OS in the atria of individual AF patients. Hypothesis: We hypothesized that in persistent AF, electrograms (EGMs) at sites of high OS, are uniquely sensitive to acute reactive oxygen species (ROS) scavenging with N-acetylcysteine (NAC), both in a large animal model and in patients and that ablation at these NAC-sensitive ‘hotspots’ can terminate or slow AF. Methods: We performed high-density mapping (both atria, pre/post-NAC, intravenous, 100mg/kg) in 17 persistent AF dogs (3-14 weeks rapid atrial pacing) using an epicardial mapping-plaque (6 regions, both atria, 130 electrodes, electrode-distance 2.5mm) (Figure 1) and in 5 patients (HD-grid, Abbott). We analyzed cycle length (CL) and dominant frequency (DF) in ‘hot spot’-sub-regions (&gt;10% change pre vs. post-NAC). In 4 dogs we performed epicardial ablation (Atricure) at 3 quadrants of the mapping fields with the largest NAC hotspots of all 6 mapped regions (Figure 2). Results: NAC led to preferential organization of AF (&gt;10% DF decrease) at several sites in both atria (NAC hot spots). In patients, NAC (Figure 3) increased CL in the left atrium in the left lateral wall and in the right atrium in the posterior lateral wall. The largest DF hotspots were in the left atrium in the superior wall. In dogs, ablation at the largest hotspots terminated AF in 50% of the cases (2/4). In all cases, ablation significantly increased CL in the left atrium and decreased DF in all atrial regions (Figure 2). Conclusions: Acute administration of NAC preferentially organizes AF electrograms in subregions of both atria (‘hotspots’). Ablation at these NAC-sensitive hotspot regions terminated or slowed AF. We have discovered a unique method to target identified atrial sub-regions of high OS in the atria. These ‘hotspots’ may be a new therapeutic target in patients with persistent AF.

Novel algorithms improve arrhythmia detection accuracy in insertable cardiac monitors.

Insertable cardiac monitors (ICMs) are commonly used to diagnose cardiac arrhythmias. False detections in the latest ICM systems remain an issue, primarily due to inaccurate R-wave sensing. New discrimination algorithms were developed and tested to reduce false detections of atrial fibrillation (AF), pause, and tachycardia episodes in ICMs. Stored electrograms (EGMs) of AF, pause, and tachycardia episodes detected by Abbott Confirm Rx™ ICMs were extracted from the Merlin.net™ Patient Care Network, and manually adjudicated to establish independent training and testing datasets. New discrimination algorithms were developed to reject false episodes due to inaccurate R-wave sensing, P-wave identification, and R-R interval patterns. The performance of these new algorithms was quantified by false positive reduction (FPR) and true positive maintenance (TPM), relative to the existing algorithms. The new AF detection algorithm was trained on 5911 EGMs from 744 devices, resulting in 66.9% FPR and 97.8% TPM. In the testing data set of 1354 EGMs from 119 devices, this algorithm achieved 45.8% FPR and 97.0% TPM. The new pause algorithm was trained on 7178 EGMs from 1490 devices, resulting in 70.9% FPR and 98.7% TPM. In the testing data set of 1442 EGMs from 340 devices, this algorithm achieved 74.4% FPR and 99.3% TPM. The new tachycardia algorithm was trained on 520 EGMs from 204 devices, resulting in 57.0% FPR and 96.6% TPM. In the testing data set of 459 EGMs from 237 devices, this algorithm achieved 57.9% FPR and 96.5% TPM. The new algorithms substantially reduced false AF, pause, and tachycardia episodes while maintaining the majority of true arrhythmia episodes detected by the Abbott ICM algorithms that exist today. Implementing these algorithms in the next-generation ICM systems may lead to improved detection accuracy, in-clinic efficiency, and device battery longevity.

Regions of Highly Recurrent Electrogram Morphology With Low Cycle Length Reflect Substrate for Atrial Fibrillation

Traditional anatomically guided ablation and attempts to perform electrogram-guided atrial fibrillation (AF) ablation (CFAE, DF, and FIRM) have not been shown to be sufficient treatment for persistent AF. Using biatrial high-density electrophysiologic mapping in a canine rapid atrial pacing model of AF, we systematically investigated the relationship of electrogram morphology recurrence (EMR) (Rec% and CLR) with established AF electrogram parameters and tissue characteristics. Rec% correlates with stability of rotational activity and with the spatial distribution of parasympathetic nerve fibers. These results have indicated that EMR may therefore be a viable therapeutic target in persistent AF.

Abstract 14782: Expression of Ngf Shrna in the Left Atrial Appendage Alone is Sufficient to Attenuate Global Left Atrial Parasympathetic and Sympathetic Nerve Sprouting and Resulting Development of Persistent Atrial Fibrillation in a Canine Model

Introduction: The autonomic nervous system plays a major role in development of atrial fibrillation (AF), with evidence of marked autonomic remodeling in a canine rapid atrial pacing (RAP) model of AF. Nerve growth factor (NGF), a neurotrophin essential for peripheral neuron growth, was upregulated in fibrillating atria, particularly in the left atrial appendage (LAA). Hypothesis: Inhibiting NGF signaling in the entire atria or the LAA alone can prevent the development of AF. Methods: NGF shRNA was injected in both atria or in the LAA of dogs followed by electroporation to facilitate gene delivery. Animals were subjected to RAP for up to 12 weeks. Duration of AF was recorded. At terminal EP study, residual AF electrograms were recorded in the posterior left atrium, left atrial free wall and LAA for analysis of AF characteristics. Tissue of each region was used for IHC with markers for parasympathetic (acetylcholinesterase) or sympathetic nerves (dopamine beta-hydroxylase). Results: Controls (n=10) developed persistent AF (&gt;8 hours) after a median of 14 days. In contrast, pan-atrial NGF shRNA animals (n=4) never developed this burden of AF. Residual AF was less fractionated (longer fractionation interval), more organized (higher organization index and lower Shannon's entropy) and more homogeneous (coefficient of variation of dominant frequency). RAP-induced nerve bundle hypertrophy and atrial myocardium hyperinnervation was significantly attenuated in pan-atrial NGF shRNA dogs. Importantly, these findings were reproduced when NGF shRNA injection was limited to the LAA alone (n=4), with prevention of development of persistent AF, and global attenuation of left atrial hyperinnervation. Conclusion: Targeted inhibition of atrial autonomic remodeling by NGF shRNA prevents development of persistent AF. NGF Inhibition in the LAA alone is sufficient to attenuate global atrial autonomic remodeling. This approach may lead to a novel, mechanism-guided therapy for AF.

Abstract 12591: Atrial Fibrillation Detection (Incidence And Accuracy) By Implantable Loop Recorders For The Indications: Cryptogenic Stroke, Syncope, Or Palpitations: A Real World Cohort Study

Introduction: Cryptogenic Stroke (CS), syncope, and palpitations are common indications for implantable loop recorder (ILR) insertion, however the incidence of new-onset atrial fibrillation (AF) in these groups in a real world clinical setting is not well characterized. Methods: At a single US center, we evaluated 273 consecutive patients who had a Medtronic Reveal LINQ™ LNQ11 implanted for the indications of CS, syncope, or palpitations from 2016-2021. All patients were followed remotely by Carelink. Patients with known AF were excluded. For each patient with an AF episode ≥ 2 minutes, the longest AF electrogram was adjudicated by an electrophysiologist and a cardiology fellow to confirm or refute the AF diagnosis. Results: Mean follow-up time was 21 months. There was no difference in mean age between the 3 groups. True, adjudicated AF was diagnosed more often in patients with palpitations and syncope than in patients with CS (table) [p &lt; 0.001]. The ILR reported AF in 92 patients (all groups), of which only 56 (61%) were adjudicated to be true AF. Of the 20 patients with a longest detected AF episode &lt;6 minutes, only 5 (25%) were true AF . Of the 72 patients with a longest detected AF of ≥ 6 minutes, 51 (71%) were true AF (p &lt; 0.001). Conclusions: True, adjudicated AF ≥ 6 minutes was detected in 11% of patients with CS, consistent with prior studies. However, in this real world cohort study, AF ≥ 6 minutes was more likely to be detected in patients with palpitations, followed by patients with syncope. The accuracy of ILR detected AF was significantly improved when the AF episode was ≥ 6 minutes duration.

Comparative Study of Methods for Cycle Length Estimation in Fractionated Electrograms of Atrial Fibrillation.

Atrial cycle length (CL) is an important feature for the analysis of electrogram (EGM) characteristics acquired during catheter ablation (CA) of atrial fibrillation (AF), the commonest cardiac arrhythmia. Nevertheless, a robust ACL estimator requires the precise detection of local activation waves (LAWs), which still remains a challenge. This work aims to compare the performance in (CL) estimation, especially under fractionated EGMs, of three different LAW detection methods relying on different operation strategies. The methods are based on the hyperbolic tangent (HT) function, an adaptive amplitude threshold (AAT) and a (CL) iteration (ACLI), respectively. For each method, LAW detection has been assessed with respect to manual annotations made by two experts and performance has been estimated by confusion matrix and mean and individual (CL) error calculation by EGM types of fractionation. The influence of EGM length on the individual (CL) error has been additionally considered. For the HT method, accuracy, sensitivity and precision were 92.77–100%, while for the AAT and ACLI methods they were 78.89–99.91% for all EGM types. The CL error on the HT method was lower than AAT and ACLI methods (up to 12 ms versus up to 20 ms), with the difference being more prominent in complex EGMs. The HT method also showed the lowest dependency on EGM length, presenting the lowest and least variable error values. Therefore, the HT method achieves higher performance in (CL) estimation in comparison with previous LAW detection techniques. The high robustness and precision demonstrated by this method suggest its implementation on CA mapping devices for a more successful location of ablation targets and improved results during CA procedures.

Safety and efficacy of catheter ablation for atrial fibrillation in the very elderly

Abstract Background The incidence and prevalence of atrial fibrillation (AF) increases with age. With an ageing general population, a 2.3-fold rise in AF prevalence is expected. Catheter ablation has emerged as an effective treatment option for rhythm control therapy. However, very elderly patients (≥80 years old) have been excluded in landmark clinical trials. Current data regarding the safety and efficacy of catheter ablation in the very elderly is therefore sparse. Purpose Due to the growing demand to manage AF in an increasingly ageing population, we investigated the safety and efficacy of catheter ablation in this particular patient population. Methods Patients with symptomatic paroxysmal, persistent and long-standing persistent AF aged ≥80 years undergoing catheter ablation, including first and re-ablation procedures in a single centre, were analysed retrospectively. Catheter ablation involved pulmonary vein isolation (PVI) using radiofrequency, cryoballoon and pulsed field ablation as energy sources. Re-ablation procedures included re-PVI and consecutive atrial tachycardia ablation including atrial lines and/or ablation of complex fractionated atrial electrograms (CFAE) in persistent AF. Endpoints included acute procedural success (complete isolation of pulmonary veins and/or non-inducibility in the case of atrial tachycardia), major complications and early arrhythmia-recurrence. Results A total of eighty-eight patients (mean age 83.1±1.9 years, mean CHA2DS2-VASc-Score 4.4±1, mean left ventricular ejection fraction 56.7±7%, direct oral anticoagulation 92.1%, vitamin-K antagonists 7.9%) were included from January 2021 to October 2021. Fifty cases (56.8%) involved PVI as an index procedure (radiofrequency 58%, n=29/50, cryoballoon 36%, n=18/50, pulsed field ablation 6%, n=3/50). Thirty-eight procedures (43.2%) involved re-ablation procedures (Re-PVI 60.5%, n=23/38, linear lesions 65.8%, n=25/38, atrial tachycardia ablation 26.3%, n=10/38 and ablation of CFAE 15.8%, n=6/38). Acute procedural success was achieved in 87/88 patients (98.9%). Major complications included stroke (n=1/88, 1.1%), pericardial tamponade (n=1/88, 1.1%) and bradycardia with subsequent pacemaker implantation (n=3/88, 3.4%). No further major complications were documented. In 13/88 patients (14.8%) early arrhythmia-recurrence occurred (38.5%, n=5/13 after the index procedure and 61.5%, n=8/13 after re-ablation) during the 90-day blanking period. Conclusions Catheter ablation for atrial fibrillation in the very elderly shows favourable acute success and low complication rates. Long term success of catheter ablation and superiority to rate control in this patient population is unknown and requires investigation in the future. Funding Acknowledgement Type of funding sources: None.

Insights From Simultaneous Left and Right Atrial Septal Mapping in Patients With Persistent Atrial Fibrillation.

The interatrial septum (IAS) is thought to be involved in the mechanism of persistent atrial fibrillation (PeAF). Simultaneous contact mapping of both sides of the IAS has not been performed previously. The purpose of this study was to describe wave front (WF) activation patterns and extent of left and right atrial septal electrical dissociation in patients with PeAF. Simultaneous mapping of both atrial septal surfaces using 2 high-density grid catheters was performed. Filtered electrograms of continuous atrial fibrillation, sinus rhythm (SR), and atrial pacing recordings were exported to MATLAB for off-line phase/activation analysis, and activation patterns on paired surfaces were analyzed. WF activation patterns between the 2 grids were evaluated to determine whether activation WFs were associated or dissociated. Eight patients with PeAF undergoing catheter ablation were included. Complete dissociation of WF activation patterns between the 2 sides of the septum existed throughout the mapping period with no 2 consecutive WF activation patterns matching. Single linear WFs were the most prevalent activation pattern on both septal grids. No focal breakthroughs were seen. Transient rotational activity was seen in 10% of phase activations. During SR and atrial pacing, both grids appeared to be activated independent of each other with no evidence of contralateral conduction across the 2 grids. Simultaneous biatrial septal mapping of human PeAF, SR, and atrial pacing shows complete WF dissociation between the left and right IAS with no evidence of trans-septal conduction, indicating that the 2 sides function as electrically discrete structures. No stable septal drivers were observed. These findings may have implications for mapping and ablation of PeAF.

An Efficient Hybrid Methodology for Local Activation Waves Detection under Complex Fractionated Atrial Electrograms of Atrial Fibrillation.

Local activation waves (LAWs) detection in complex fractionated atrial electrograms (CFAEs) during catheter ablation (CA) of atrial fibrillation (AF), the commonest cardiac arrhythmia, is a complicated task due to their extreme variability and heterogeneity in amplitude and morphology. There are few published works on reliable LAWs detectors, which are efficient for regular or low fractionated bipolar electrograms (EGMs) but lack satisfactory results when CFAEs are analyzed. The aim of the present work is the development of a novel optimized method for LAWs detection in CFAEs in order to assist cardiac mapping and catheter ablation (CA) guidance. The database consists of 119 bipolar EGMs classified by AF types according to Wells’ classification. The proposed method introduces an alternative Botteron’s preprocessing technique targeting the slow and small-ampitude activations. The lower band-pass filter cut-off frequency is modified to 20 Hz, and a hyperbolic tangent function is applied over CFAEs. Detection is firstly performed through an amplitude-based threshold and an escalating cycle-length (CL) analysis. Activation time is calculated at each LAW’s barycenter. Analysis is applied in five-second overlapping segments. LAWs were manually annotated by two experts and compared with algorithm-annotated LAWs. AF types I and II showed accuracy and sensitivity. AF type III showed accuracy and sensitivity. The results of this study highlight the efficiency of the developed method in precisely detecting LAWs in CFAEs. Hence, it could be implemented on real-time mapping devices and used during CA, providing robust detection results regardless of the fractionation degree of the analyzed recordings.

PO-617-05 PREDICTING ABLATION RESPONSE IN PERSISTENT ATRIAL FIBRILLATION BY MACHINE LEARNING OF GLOBAL ATRIAL ELECTROGRAM PATTERNS

It remains unclear if patterns of electrical activity in patients with persistent atrial fibrillation (AF) predict clinical endpoints. We hypothesized that machine learning (ML) of AF electrogram patterns (EGM) can predict acute termination by ablation.

Endomysial fibrosis, rather than overall connective tissue content, is the main determinant of conduction disturbances in human atrial fibrillation.

AimsAlthough in persistent atrial fibrillation (AF) a complex AF substrate characterized by a high incidence of conduction block has been reported, relatively little is known about AF complexity in paroxysmal AF (pAF). Also, the relative contribution of various aspects of structural alterations to conduction disturbances is not clear. In particular, the contribution of endomysial fibrosis to conduction disturbances during progression of AF has not been studied yet.Methods and resultsDuring cardiac surgery, epicardial high-density mapping was performed in patients with acutely induced (aAF, n = 11), pAF (n = 12), and longstanding persistent AF (persAF, n = 9) on the right atrial (RA) wall, the posterior left atrial wall (pLA) and the LA appendage (LAA). In RA appendages, overall and endomysial (myocyte-to-myocyte distances) fibrosis and connexin 43 (Cx43) distribution were quantified. Unipolar AF electrogram analysis showed a more complex pattern with a larger number of narrower waves, more breakthroughs and a higher fractionation index (FI) in persAF compared with aAF and pAF, with no differences between aAF and pAF. The FI was consistently higher at the pLA compared with the RA. Structurally, Cx43 lateralization increased with AF progression (aAF = 7.5 ± 8.9%, pAF = 24.7 ± 11.1%, persAF = 35.1 ± 11.4%, P < 0.001). Endomysial but not overall fibrosis correlated with AF complexity (r = 0.57, P = 0.001; r = 0.23, P = 0.20; respectively).ConclusionsAtrial fibrillation complexity is highly variable in patients with pAF, but not significantly higher than in patients with acutely induced AF, while in patients with persistent AF complexity is higher. Among the structural alterations studied, endomysial fibrosis, but not overall fibrosis, is the strongest determinant of AF complexity.

Abstract 14297: Spatially Synchronized Electrogram Islands Within Atrial Fibrillation Predict Termination by Ablation

Introduction: It is unclear how persistent atrial fibrillation (AF) terminates by any spatial arrangement of ablation lesions. Hypothesis: We hypothesized that AF that exhibits spatial organization with islands of 1:1 synchronized electrograms within the disordered milieu and, specifically, that patients with larger area islands may be more likely to terminate than those without. Methods: We recruited n=60 patients (72% male, age 63±10 years) in whom AF terminated by ablation (n=30; “Term”) or did not (n=30, “Non-term”). Unipolar AF electrograms were recorded by 64-pole basket catheters in both atria. For all possible 2X2 electrode grids, synchronization was assessed by correlation (repeating shapes + relative timings, coefficient, Fig A). Ablation targeted organized sites (rotational/ focal), then pulmonary vein isolation. Results: Fig A shows (top) AF EGMs in a Term patient showing 1:1 patterns at D2-D3-E2-E3 over time (correlation&gt;0.9). For the entire left atrium, there was a large synchronized island &gt;25% of mapped atria (Fig B, red). Bottom panels show less synchronization in a Non-term patient (Fig A), with &lt;5% area (Fig B). For all patients, synchronized islands (with correlation &gt;0.8) were larger for Term (black) than Non-term (gray) patients over a range of time durations (Fig C, p&lt;0.01 ANOVA). Area of synchronized regions predicted AF termination with c-statistic 0.68, and the optimum cut-point provided sensitivity 87% (specificity 43%). Conclusions: Persistent AF shows organized islands of synchronized electrograms, whose size and duration predict the likelihood of termination by ablation. Future studies should investigate the pathophysiology of synchronized islands in AF.

Unveiling the impact of fibrosis presence in fibrillatory electrograms

Abstract Introduction Complexity and fragmentation presence in atrial fibrillation (AF) electrograms (EGMs) has been extensively studied in the literature, with areas exhibiting complex fractionated atrial electrograms (CFAEs) postulated as responsible for AF maintenance. Computational simulations based on in silico models are widely employed to support AF initiating and maintenance theories. However, they are still limited when trying to mimic the effect of fibrosis in live cardiac cells and its impact in the clinical context, i.e., EGM morphologies in electroanatomical mapping studies. Controlled experiments on cultivated cardiomyocytes and optical mapping signals can better explain the role of fibrosis and its effect on EGM recordings. Purpose To study the effect of fibrosis in fibrillatory scenarios by comparing the temporal and frequency domains of EGMs and Ca2+ transient signals in cardiomyocyte cell cultures. Methods A total of 51 HL1 cardiomyocytes cell line monolayers experiments were cultured until confluence. Optical mapping (OM) was performed to evaluate the electrophysiological activity of the samples through Ca2+ transient quantification under the effect of Rhod2 dye. Activation frequency and calcium transient signals were acquired, and virtual EGMs were calculated from the signals virtually after post-processing the cell cultures at 16 different positions in the culture, see Figure 1A. Results The DF ratio between the OM and the virtual EGMs obtained for the different fibrosis levels are summarized and represented in Figure 1B. The Pearson's correlation coefficient was adjusted for all the samples and the significance level was stablished at p-value&amp;lt;0.01. Conclusions Optical mapping signal morphologies strongly correlate with EGM recordings when a clear dominant frequency (DF) dominates the complete area of the cell culture and no fibrosis is present. However, for patchy distributions of the DF, usually associated with higher concentrations of fibrosis, morphologies do not correlate for increasing fibrosis levels, and DF measurements differ with poor match between EGM-DF and OM-DF. Results on higher fibrotic experiments, i.e., 40–50% fibrosis level, show that DF correlates due to the poor and degraded cardiac activity in the EGMs and optical mapping recordings. The results reveal the importance and impact of fibrosis in the remodeling of atrial tissue and the correct interpretation of endocardial EGMs in the presence of heterogeneous tissue. Funding Acknowledgement Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Instituto de Salud Carlos III and Ministerio de Ciencia, Innovaciόn y Universidades

B-PO04-062 COMPARING THE ACCURACY OF FEATURES TO IDENTIFY ATRIAL FIBRILLATION ON INTRACARDIAC ELECTROGRAMS

Cardiac devices are challenged when separating atrial fibrillation (AF) from other tachycardias (AT) from single electrograms (EGM). For this reason, devices tend to group arrhythmias by a rate threshold. We set out to determine the accuracy of multiple clinically-intuitive EGM features to separate AF from AT, using a deep learning (DL) logic. In 86 patients (61 male, age 65±11) we obtained a total of 29,340 EGMs of AF (4 s) from intracardiac channels (fig A) and AT pre-ablation. EGM variability was seen in both arrhythmias. We comprehensively probed EGM using 49 clinically-intuitive features based on cycle length, frequency and amplitude (fig B), analysed individually and combined. We developed custom convolutional (CNN) and recurrent (RNN) neural networks. Classifiers were tested using the same 10-fold cross-validation folds (80% patients for training, 20% for testing). Single EGM features modestly separated AF from AT with c-statistics ranging from 0.55 (rate) to 0.82 (autocorrelation of EGM shape). Accuracy increased when features were combined (fig C) Linearly (c-statistic 0.95 ± 0.04), by Bagged Trees (0.95 ± 0.04), K-Nearest Neighbors (0.95 ± 0.04) or Support Vector Machines (0.95 ± 0.04). DL on raw EGMs provided higher accuracy without the need to calculate features, with c-statistics up to 0.97 ± 0.04 (RNN; fig C). Traditional EGM features separate AF from AT, but ‘high rate events’ as a sole criterion is poor. Combining shape and regularity indices provides the optimal approach. This limitation must be considered in future approaches to automatically detect AF from devices to guide clinical care.

Deep Learning Classification of Unipolar Electrograms in Human Atrial Fibrillation: Application in Focal Source Mapping.

Focal sources are potential targets for atrial fibrillation (AF) catheter ablation, but they can be time-consuming and challenging to identify when unipolar electrograms (EGM) are numerous and complex. Our aim was to apply deep learning (DL) to raw unipolar EGMs in order to automate putative focal sources detection. We included 78 patients from the Focal Source and Trigger (FaST) randomized controlled trial that evaluated the efficacy of adjunctive FaST ablation compared to pulmonary vein isolation alone in reducing AF recurrence. FaST sites were identified based on manual classification of sustained periodic unipolar QS EGMs over 5-s. All periodic unipolar EGMs were divided into training (n = 10,004) and testing cohorts (n = 3,180). DL was developed using residual convolutional neural network to discriminate between FaST and non-FaST. A gradient-based method was applied to interpret the DL model. DL classified FaST with a receiver operator characteristic area under curve of 0.904 ± 0.010 (cross-validation) and 0.923 ± 0.003 (testing). At a prespecified sensitivity of 90%, the specificity and accuracy were 81.9 and 82.5%, respectively, in detecting FaST. DL had similar performance (sensitivity 78%, specificity 89%) to that of FaST re-classification by cardiologists (sensitivity 78%, specificity 79%). The gradient-based interpretation demonstrated accurate tracking of unipolar QS complexes by select DL convolutional layers. In conclusion, our novel DL model trained on raw unipolar EGMs allowed automated and accurate classification of FaST sites. Performance was similar to FaST re-classification by cardiologists. Future application of DL to classify FaST may improve the efficiency of real-time focal source detection for targeted AF ablation therapy.

B-PO02-044 ELECTROGRAM FINGERPRINTS OF ATRIAL FIBRILLATION

It is unclear whether disorganized waves in atrial fibrillation (AF) produce distinct ‘signature shapes’ on the electrogram (EGM), reflecting specific patterns. We tested the hypothesis that deep learning (DL) could be used to identify specific EGM shapes pathognomic for AF, that could form the basis for future approaches to detect subtype and treat patients. We developed custom-built convolutional (CNN) and recurrent (RNN) neural networks using N=29,340 EGM segments (4 sec), from 86 patients with AF or AFL at ablation (25 female, 65±11 years), and using 10-fold cross-validation (80% patients for training, 20% for independent testing). We probed CNN by generating 29,880 EGM sequences from patient data, in which we successively controlled variables including cycle length (CL) and EGM shape. Trained CNN and RNN identified AF from other atrial arrhythmias with an AUC of 0.95 ± 0.05 and 0.97 ± 0.04 in the testing sets respectively (Fig. A). We identified EGM shapes that were classified as AF even if they had regular rhythm and slow rate (long CL). A systematic analysis of 101 such AF signature EGMs were classified as AF >80% by CNN regardless of rate and irregularity (fig B, orange), and also by the RNN on 30 of such EGM signatures. Signature AF electrograms were identical by both RNN and CNN classification for 15 shapes (fig C, red shapes). Deep learning identified signature AF electrograms. EGM signature shapes were more important to AF L classification than high rate or irregular timing. Extraction of complex electrophysiological information hidden to the expert eye may improve diagnosis and shed insights into mechanisms of EGM generation in AF.

Incidence, prevalence, and trajectories of repetitive conduction patterns in human atrial fibrillation.

Repetitive conduction patterns in atrial fibrillation (AF) may reflect anatomical structures harbouring preferential conduction paths and indicate the presence of stationary sources for AF. Recently, we demonstrated a novel technique to detect repetitive patterns in high-density contact mapping of AF. As a first step towards repetitive pattern mapping to guide AF ablation, we determined the incidence, prevalence, and trajectories of repetitive conduction patterns in epicardial contact mapping of paroxysmal and persistent AF patients. A 256-channel mapping array was used to record epicardial left and right AF electrograms in persistent AF (persAF, n = 9) and paroxysmal AF (pAF, n = 11) patients. Intervals containing repetitive conduction patterns were detected using recurrence plots. Activation movies, preferential conduction direction, and average activation sequence were used to characterize and classify conduction patterns. Repetitive patterns were identified in 33/40 recordings. Repetitive patterns were more prevalent in pAF compared with persAF [pAF: median 59%, inter-quartile range (41-72) vs. persAF: 39% (0-51), P < 0.01], larger [pAF: = 1.54 (1.15-1.96) vs. persAF: 1.16 (0.74-1.56) cm2, P < 0.001), and more stable [normalized preferentiality (0-1) pAF: 0.38 (0.25-0.50) vs. persAF: 0.23 (0-0.33), P < 0.01]. Most repetitive patterns were peripheral waves (87%), often with conduction block (69%), while breakthroughs (9%) and re-entries (2%) occurred less frequently. High-density epicardial contact mapping in AF patients reveals frequent repetitive conduction patterns. In persistent AF patients, repetitive patterns were less frequent, smaller, and more variable than in paroxysmal AF patients. Future research should elucidate whether these patterns can help in finding AF ablation targets.