Search

BACKGROUND: Accurately determining arrhythmia mechanism from a 12-lead electrocardiogram (ECG) of supraventricular tachycardia can be challenging. We hypothesized a convolutional neural network (CNN) can be trained to classify atrioventricular re-entrant tachycardia (AVRT) vs atrioventricular nodal re-entrant tachycardia (AVNRT) from the 12-lead ECG, when using findings from the invasive electrophysiology (EP) study as the gold standard. METHODS: We trained a CNN on data from 124 patients undergoing EP studies with a final diagnosis of AVRT or AVNRT. A total of 4962 5-second 12-lead ECG segments were used for training. Each case was labeled AVRT or AVNRT based on the findings of the EP study. The model performance was evaluated against a hold-out test set of 31 patients and compared to an existing manual algorithm. RESULTS: The model had an accuracy of 77.4% in distinguishing between AVRT and AVNRT. The area under the receiver operating characteristic curve was 0.80. In comparison, the existing manual algorithm achieved an accuracy of 67.7% on the same test set. Saliency mapping demonstrated the network used the expected sections of the ECGs for diagnoses; these were the QRS complexes that may contain retrograde P waves. CONCLUSION: We describe the first neural network trained to differentiate AVRT from AVNRT. Accurate diagnosis of arrhythmia mechanism from a 12-lead ECG could aid preprocedural counseling, consent, and procedure planning. The current accuracy from our neural network is modest but may be improved with a larger training dataset.

Aims Accurately determining atrial arrhythmia mechanisms from a 12-lead electrocardiogram (ECG) can be challenging. Given the high success rate of cavotricuspid isthmus (CTI) ablation, identification of CTI-dependent typical atrial flutter (AFL) is important for treatment decisions and procedure planning. We sought to train a convolutional neural network (CNN) to classify CTI-dependent AFL vs. non-CTI dependent atrial tachycardia (AT), using data from the invasive electrophysiology (EP) study as the gold standard. Methods and results We trained a CNN on data from 231 patients undergoing EP studies for atrial tachyarrhythmia. A total of 13 500 five-second 12-lead ECG segments were used for training. Each case was labelled CTI-dependent AFL or non-CTI-dependent AT based on the findings of the EP study. The model performance was evaluated against a test set of 57 patients. A survey of electrophysiologists in Europe was undertaken on the same 57 ECGs. The model had an accuracy of 86% (95% CI 0.77–0.95) compared to median expert electrophysiologist accuracy of 79% (range 70–84%). In the two thirds of test set cases (38/57) where both the model and electrophysiologist consensus were in agreement, the prediction accuracy was 100%. Saliency mapping demonstrated atrial activation was the most important segment of the ECG for determining model output. Conclusion We describe the first CNN trained to differentiate CTI-dependent AFL from other AT using the ECG. Our model matched and complemented expert electrophysiologist performance. Automated artificial intelligence-enhanced ECG analysis could help guide treatment decisions and plan ablation procedures for patients with organized atrial arrhythmias.

Micro-anatomical reentry has been identified as a potential driver of atrial fibrillation (AF). In this paper, we introduce a novel computational method which aims to identify which atrial regions are most susceptible to micro-reentry. The approach, which considers the structural basis for micro-reentry only, is based on the premise that the accumulation of electrically insulating interstitial fibrosis can be modelled by simulating percolation-like phenomena on spatial networks. Our results suggest that at high coupling, where micro-reentry is rare, the micro-reentrant substrate is highly clustered in areas where the atrial walls are thin and have convex wall morphology. However, as transverse connections between fibres are removed, mimicking the accumulation of interstitial fibrosis, the substrate becomes less spatially clustered, and the bias to forming in thin, convex regions of the atria is reduced. Comparing our algorithm on image-based models with and without atrial fibre structure, we find that strong longitudinal fibre coupling can suppress the micro-reentrant substrate, whereas regions with disordered fibre orientations have an enhanced risk of micro-reentry. We suggest that with further development, these methods may have future potential for patient-specific risk stratification, taking a longitudinal view of the development of the micro-reentrant substrate.Author summaryAtrial fibrillation (AF) is the most common abnormal heart rhythm, yet, despite extensive research, treatment success rates remain poor. In part, this is because there is an incomplete understanding of the mechanistic origin of AF. In this paper, we investigate one proposed mechanism of AF, the formation of “micro-reentrant circuits”, which can be thought of as a “short circuit”, forming when electrically insulating fibrosis (structural repair tissue) infiltrates the space between heart muscle cells. Previously, such circuits have been found in experimental hearts, but identifying these circuits clinically is difficult. Here, we aim to take a small step towards developing computational methods for identifying where in the atria these circuits are most likely to form, drawing on techniques from network science. Our approach indicates that a number of factors are key to determining where circuits form, most notably the thickness of the heart muscle, and the alignment of muscle fibres.

Background: Atrial fibrillation (AF) and ventricular fibrillation (VF) are complex heart rhythm disorders and may be sustained by distinct electrophysiological mechanisms. Disorganised self-perpetuating multiple-wavelets and organised rotational drivers (RDs) localising to specific areas are both possible mechanisms by which fibrillation is sustained. Determining the underlying mechanisms of fibrillation may be helpful in tailoring treatment strategies. We investigated whether global fibrillation organisation, a surrogate for fibrillation mechanism, can be determined from electrocardiograms (ECGs) using band-power (BP) feature analysis and machine learning.Methods: In this study, we proposed a novel ECG classification framework to differentiate fibrillation organisation levels. BP features were derived from surface ECGs and fed to a linear discriminant analysis classifier to predict fibrillation organisation level. Two datasets, single-channel ECGs of rat VF (n = 9) and 12-lead ECGs of human AF (n = 17), were used for model evaluation in a leave-one-out (LOO) manner.Results: The proposed method correctly predicted the organisation level from rat VF ECG with the sensitivity of 75%, specificity of 80%, and accuracy of 78%, and from clinical AF ECG with the sensitivity of 80%, specificity of 92%, and accuracy of 88%.Conclusion: Our proposed method can distinguish between AF/VF of different global organisation levels non-invasively from the ECG alone. This may aid in patient selection and guiding mechanism-directed tailored treatment strategies.

INTRODUCTION: We hypothesized that very high density mapping of typical atrial flutter (AFL) would facilitate a more complete understanding of its circuit. Such very high density mapping was performed with the Rhythmia mapping system using its 64 electrode basket catheter. METHODS AND RESULTS: Data were acquired from 13 patients in AFL. Functional anatomy of the right atrium (RA) was readily identified during mapping including the Crista Terminalis and Eustachian ridge. The leading edge of the activation wavefront was identified without interruption and its conduction velocity (CV) calculated. CV was not different at the cavotricuspid isthmus (CTI) compared to the remainder of the RA (1.02 vs. 1.03 m/s, p = 0.93). The sawtooth pattern of the surface EKG flutter waves were compared to the position of the dominant wavefront. The downslope of the surface EKG flutter waves represented on average, 73% ± 9% of the total flutter cycle length. During the downslope the activation wavefront travelled significantly further than during the upslope (182 ± 21 ms vs. 68 ± 29 ms, p

AimsAblation of persistent atrial fibrillation (PsAF) has been performed by many techniques with varying success rates. This may be due to ablation techniques, patient demographics, comorbidities, and trial design. We conducted a meta-regression of studies of PsAF ablation to elucidate the factors affecting atrial fibrillation (AF) recurrence.Methods and results Databases were searched for prospective studies of PsAF ablation. A meta-regression was performed. Fifty-eight studies (6767 patients) were included. Complex fractionated atrial electrogram (CFAE) ablation reduced freedom from AF by 8.9% [95% confidence interval (CI) −15 to −2.3, P = 0.009). Left atrial appendage [LAA isolation (three study arms)] increased freedom from AF by 39.5% (95% CI 9.1–78.4, P = 0.008). Posterior wall isolation (PWI) (eight study arms) increased freedom from AF by 19.4% (95% CI 3.3–38.1, P = 0.017). Linear ablation or ganglionated plexi ablation resulted in no significant effect on freedom from AF. More extensive ablation increased intraprocedural AF termination; however, intraprocedural AF termination was not associated with improved outcomes. Increased left atrial diameter was associated with a reduction in freedom from AF by 4% (95% CI −6.8% to −1.1%, P = 0.007) for every 1 mm increase in diameter.Conclusion Linear ablation, PWI, and CFAE ablation improves intraprocedural AF termination, but such termination does not predict better long-term outcomes. Study arms including PWI or LAA isolation in the lesion set were associated with improved outcomes in terms of freedom from AF; however, further randomized trials are required before these can be routinely recommended. Left atrial size is the most important marker of AF chronicity influencing outcomes.

Aims The aim of this study was to describe the head-up tilt (HUT) test and carotid sinus massage (CSM) responses, and the occurrence of syncope with coughing during HUT in a large cohort of patients. Methods and results A total of 5133 HUT were retrospectively analysed to identify patients with cough syncope. Head-up tilt followed by CSM were performed. Patients were made to cough on two separate occasions in an attempt to reproduce typical clinical symptoms on HUT. Patients with cough syncope were compared with 29 age-matched control patients with syncope unrelated to coughing. A total of 29 patients (26 male, age 49 ± 14 years) with cough syncope were identified. Coughing during HUT reproduced typical prodromal symptoms of syncope in 16 (55%) patients and complete loss of consciousness in 2 (7%) patients, with a mean systolic blood pressure reduction of 45 ± 26 mmHg, and a mean increase in heart rate of 13 ± 8 b.p.m. No syncope or symptoms after coughing were observed in the control group. The HUT result was positive in 13 (48%) patients with the majority of positive HUT responses being vasodepressor (70% of positive HUT). Carotid sinus massage was performed in 18 patients being positive with a vasodepressor response causing mild pre-syncopal symptoms in only 1 patient. Conclusion Syncope during coughing is a result of hypotension, rather than bradycardia. Coughing during HUT is a useful test in patients suspected to have cough syncope but in whom the history is not conclusive.

Transcatheter aortic valve implantation (TAVI) is an effective treatment option for patients with severe degenerative aortic valve stenosis who are high risk for conventional surgery. Computed tomography (CT) performed prior to TAVI can detect pathologies that could influence outcomes following the procedure, however the incidence, cost, and clinical impact of incidental findings has not previously been investigated. 279 patients underwent CT; 188 subsequently had TAVI and 91 were declined. Incidental findings were classified as clinically significant (requiring treatment), indeterminate (requiring further assessment), or clinically insignificant. The primary outcome measure was all-cause mortality up to 3 years. Costs incurred by additional investigations resultant to incidental findings were estimated using the UK Department of Health Payment Tariff. Incidental findings were common in both the TAVI and medical therapy cohorts (54.8 vs. 70.3 %; P = 0.014). Subsequently, 45 extra investigations were recommended for the TAVI cohort, at an overall average cost of £32.69 per TAVI patient. In a univariate model, survival was significantly associated with the presence of a clinically significant or indeterminate finding (HR 1.61; P = 0.021). However, on multivariate analysis outcomes after TAVI were not influenced by any category of incidental finding. Incidental findings are common on CT scans performed prior to TAVI. However, the total cost involved in investigating these findings is low, and incidental findings do not independently identify patients with poorer outcomes after TAVI. The discovery of an incidental finding on CT should not necessarily influence or delay the decision to perform TAVI.

Aims Full-disclosure study describing Doppler patterns during iterative atrioventricular delay (AVD) optimization of biventricular pacemakers (cardiac resynchronization therapy, CRT). Method and results Doppler traces of the first 50 eligible patients undergoing iterative Doppler AVD optimization in the BRAVO trial were examined. Three experienced observers classified conformity to guideline-described patterns. Each observer then selected the optimum AVD on two separate occasions: blinded and unblinded to AVD. Four Doppler E-A patterns occurred: A (always merged, 18% of patients), B (incrementally less fusion at short AVDs, 12%), C (full separation at short AVDs, as described by the guidelines, 28%), and D (always separated, 42%). In Groups A and D (60%), the iterative guidelines therefore cannot specify one single AVD. On the kappa scale (0 = chance alone; 1 = perfect agreement), observer agreement for the ideal AVD in Classes B and C was poor (0.32) and appeared worse in Groups A and D (0.22). Blinding caused the scattering of the AVD selected as optimal to widen (standard deviation rising from 37 to 49 ms, P < 0.001). By blinding 28% of the selected optimum AVDs were ≤60 or ≥200 ms. All 50 Doppler datasets are presented, to support future methodological testing. Conclusion In most patients, the iterative method does not clearly specify one AVD. In all the patients, agreement on the ideal AVD between skilled observers viewing identical images is poor. The iterative protocol may successfully exclude some extremely unsuitable AVDs, but so might simply accepting factory default. Irreproducibility of the gold standard also prevents alternative physiological optimization methods from being validated honestly.

Electroanatomical mapping technology has developed rapidly over recent years. The Rhythmia mapping system (Boston Scientific, Natick, Massachusetts) has been described in detail elsewhere [(1)][1]. Briefly the system involves automated collection of anatomical and electrogram data from the atrium

Background Head-up tilt (HUT) is used for diagnosis of vasovagal syncope (VVS), and can provoke cardioinhibition. VVS is usually considered benign, however pacemaker insertion may be indicated in some patients. We sought to characterize the long-term outcomes of patients with prolonged asystole (>15s) on HUT. Methods We conducted a retrospective study on patients with asystole >15s on HUT identified from 5133 patients who were investigated between 1998 and 2012 at our institution. Patients were mailed questionnaires or telephoned to ascertain outcomes. Where contact was not possible, the patients' general practitioners were contacted to request up-to-date information. Results A total of 26 patients with a mean age of 45±18years and a mean duration of asystole on HUT of 26±7s were successfully followed up from a total of 77 patients identified. The follow-up duration was 99±39months. Six patients had undergone pacemaker (PPM) implantation. Of the patients without PPM, 16 reported spontaneously improved symptoms. Ten patients sustained injury prior to HUT compared with one after HUT, when a clear diagnosis was made and management advice was given. There were no major injuries or deaths after HUT. The 6 patients with PPMs had a mean age of 60±16 (67% male) at HUT. Four patients had no further syncope after PPM and two demonstrated improvement but still experienced recurrent syncope. Conclusions Prolonged asystole (>15s) on tilt does not necessarily predict adverse outcomes with most patients improving spontaneously over the long-term. Pacemaker insertion in selected patients may reduce syncope recurrence but does not always abolish it.

Syncope is a common symptom with many causes. Affecting a large proportion of the population, both young and old, it represents a significant healthcare burden. The diagnostic approach to syncope should be focused on the initial evaluation, which includes a detailed clinical history, physical examination and 12-lead electrocardiogram. Following the initial evaluation, patients should be risk-stratified into high or low-risk groups in order to guide further investigations and management. Patients with high-risk features should be investigated further to exclude significant structural heart disease or arrhythmia. The ideal currently-available investigation should allow ECG recording during a spontaneous episode of syncope, and when this is not possible, an implantable loop recorder may be considered. In the emergency room setting, acute causes of syncope must also be considered including severe cardiovascular compromise due to pulmonary, cardiac or vascular pathology. While not all patients will receive a conclusive diagnosis, risk-stratification in patients to guide appropriate investigations in the context of a diagnostic algorithm should allow a benign prognosis to be maintained.

Background The European Society of Cardiology’s guidelines “present all the relevant evidence on a particular clinical issue” to clinicians, being designed to be “helpful in everyday clinical medical decision-making”.Statements are accompanied by one of three Levels of Evidence: A (multiple randomised controlled trials (RCTs) or meta-analyses), B (large non-randomised studies or one RCT) or C (expert opinion, small/retrospective studies or registry data). We evaluated the Levels of evidence underlying the guideline documents as a whole to establish whether, as time progresses, they are providing clinicians with recommendations of increasing scientific soundness. Methods We systematically analysed all current and past guideline documents from the European Society of Cardiology, which offered recommendations with Levels of Evidence or Class of Recommendation.We recorded the Level of Evidence available for each recommendation. Two independent observers extracted the Level of Evidence provided for each recommendation.A third observer settled any disagreements. Results There are now over 3000 active guideline recommendations to steer the practice of cardiologists. Almost half (49.6%) of recommendations are Level C and are based on no more than opinion or bias-vulnerable observational reports.Under a fifth (18.3%) have the highest recommendation level, A and are based on randomised controlled trials or meta-analyses.Level B accounts for the remaining 32.1%. There has been a large increase in the number of guideline recommendations over time (Figure 1). Revision of an individual guideline document increases the total number of recommendations (median of +7.5, IQR 0 to +49).Recommendations based on the lowest Level of Evidence (C) were more frequently added (median of +17.5, IQR 0 to +36) than Level B (+5.5, IQR -0.5 to +14.75), which were in turn more commonly added than recommendations with the highest Level of Evidence (A) (+3.5, IQR 0 to +11). Areas of cardiology differed substantially in Levels of Evidence. ACE inhibitors had the highest proportion of high-level evidence (86% Level A, 7% B and 7% C) whilst Aortic Dissection the lowest (100% Level C). Over 2/3 of guideline documents had less than 1/3 of their recommendations based on Level A.Only 2 had over 50% of recommendations based on Level A. Conclusions Cardiologists are faced with over 3000 evidence-based instructions to steer their clinical practice.Almost half have no RCT data whatsoever.Recommendations continue to be added, but these are most commonly based on opinion, not RCT proof. We recommend that if guidelines are to direct clinical action they should restrain themselves to making a smaller number of clear recommendations that are based on reliable (i.e. RCT) evidence.