Inter- and intra-observer variability of visual fragmented QRS scoring in ischemic and non-ischemic cardiomyopathy

Life-threatening ventricular arrhythmia prediction in patients with dilated cardiomyopathy using explainable electrocardiogram-based deep neural networks

AIMS

While electrocardiogram (ECG) characteristics have been associated with life-threatening ventricular arrhythmias (LTVA) in dilated cardiomyopathy (DCM), they typically rely on human-derived parameters. Deep neural networks (DNNs) can discover complex ECG patterns, but the interpretation is hampered by their 'black-box' characteristics. We aimed to detect DCM patients at risk of LTVA using an inherently explainable DNN.

METHODS AND RESULTS

In this two-phase study, we first developed a variational autoencoder DNN on more than 1 million 12-lead median beat ECGs, compressing the ECG into 21 different factors (F): FactorECG. Next, we used two cohorts with a combined total of 695 DCM patients and entered these factors in a Cox regression for the composite LTVA outcome, which was defined as sudden cardiac arrest, spontaneous sustained ventricular tachycardia, or implantable cardioverter-defibrillator treated ventricular arrhythmia. Most patients were male (n = 442, 64%) with a median age of 54 years [interquartile range (IQR) 44-62], and median left ventricular ejection fraction of 30% (IQR 23-39). A total of 115 patients (16.5%) reached the study outcome. Factors F8 (prolonged PR-interval and P-wave duration, P < 0.005), F15 (reduced P-wave height, P = 0.04), F25 (increased right bundle branch delay, P = 0.02), F27 (P-wave axis P < 0.005), and F32 (reduced QRS-T voltages P = 0.03) were significantly associated with LTVA.

CONCLUSION

Inherently explainable DNNs can detect patients at risk of LTVA which is mainly driven by P-wave abnormalities.

Validation of CalECG software for primary prevention heart failure patients: Reducing inter-observer measurement variability

BACKGROUND

In primary prevention heart failure patients the 12‑lead electrocardiogram (ECG) may be useful for the prediction of ventricular arrhythmias. However, inter-observer measurement variability first needs to be identified and any software used, validated.

OBJECTIVE

To compare manual ECG measures with CalECG software and to assess the reliability of visual recognition of fragmented QRS (fQRS) by advanced cardiology trainees.

METHODS

30 pre-implant ECGs were assessed on patients who met guidelines for primary prevention Implantable Cardiac Defibrillator. Parameters included RR, PR, QT, QRS duration, axis location, fQRS and T wave peak to T wave end (TpTe). ECGs were analyzed by members of the cardiology department with different levels of experience, and compared to CalECG software. Interobserver agreement was assessed using Fleiss' Kappa (κ) and intraclass correlation coefficients (ICC). Pearson correlation coefficient (r) was used to compare human and software measures.

RESULTS

Strong/very strong correlation was recorded across manual ECG measures (ICC = 0.749-0.979, p ≤ 0.0001) with moderate/strong correlation for TpTe (ICC = 0.547-0.765, p ≤ 0.001). Advanced cardiology trainees demonstrated substantial agreement on ECG interpretation (κ = 0.788, p ≤ 0.0001), however, reliability of fQRS assessment was only moderate for identification (κ = 0.5, p ≤ 0.0001) and fair for location (κ = 0.295, p = 0.001). CalECG software showed strong/very strong correlation with manual measurement for standard measures (r = 0.756-0.977, p ≤ 0.001). Concordance between human and software TpTe measurements varied between leads, with V5 showing a non-significant weak correlation (r = 0.197).

CONCLUSION

CalECG software showed strong/very strong correlation with standard manual measures which affirms its use in ECG analysis. Advanced cardiology trainees showed greater variability in the identification and location of fQRS.

A machine learning algorithm for electrocardiographic fQRS quantification validated on multi-center data

Fragmented QRS (fQRS) is an electrocardiographic (ECG) marker of myocardial conduction abnormality, characterized by additional notches in the QRS complex. The presence of fQRS has been associated with an increased risk of all-cause mortality and arrhythmia in patients with cardiovascular disease. However, current binary visual analysis is prone to intra- and inter-observer variability and different definitions are problematic in clinical practice. Therefore, objective quantification of fQRS is needed and could further improve risk stratification of these patients. We present an automated method for fQRS detection and quantification. First, a novel robust QRS complex segmentation strategy is proposed, which combines multi-lead information and excludes abnormal heartbeats automatically. Afterwards extracted features, based on variational mode decomposition (VMD), phase-rectified signal averaging (PRSA) and the number of baseline-crossings of the ECG, were used to train a machine learning classifier (Support Vector Machine) to discriminate fragmented from non-fragmented ECG-traces using multi-center data and combining different fQRS criteria used in clinical settings. The best model was trained on the combination of two independent previously annotated datasets and, compared to these visual fQRS annotations, achieved Kappa scores of 0.68 and 0.44, respectively. We also show that the algorithm might be used in both regular sinus rhythm and irregular beats during atrial fibrillation. These results demonstrate that the proposed approach could be relevant for clinical practice by objectively assessing and quantifying fQRS. The study sets the path for further clinical application of the developed automated fQRS algorithm.

Repolarization abnormalities on admission predict 1-year outcome in COVID-19 patients

Background

ECG abnormalities in COVID-19 have been widely reported, however data after discharge is limited. The aim was to describe ECG abnormalities on admission and following recovery of COVID-19, and their associated mortality.

Methods

All patients hospitalized in a tertiary care hospital between March 7th and July 1st 2020 with COVID-19 were included in a retrospective registry. The first ECG on admission was collected, together with an ECG after hospital discharge in the absence of acute pathology. Automated measures and clinical ECG interpretations were collected. Multivariate Cox regression analysis was performed to predict 1-year all-cause mortality.

Results

In total 420 patients were included, of which 83 patients (19.8%) died during the 1-year follow-up period. Repolarization abnormalities were present in 189 patients (45.0%). The extent of repolarization abnormalities was an independent predictor of 1-year all-cause mortality (HR per region 1.30, 95%CI 1.04–1.64) together with age (/year HR 1.06, 95%CI 1.04–1.08), heart rate (/bpm HR 1.02, 95%CI 1.01–1.03), neurological disorders (HR 2.41, 95%CI 1.47–3.93), active cancer (HR 2.75, 95%CI 1.57–4.82), CRP (per 10 mg/L HR 1.05, 95%CI 1.02–1.08) and eGFR (per 10 mg/L HR 0.90, 95%CI 0.83–0.98).

In 245 patients (68.1%) an ECG post discharge was available. New repolarization abnormalities were more frequent in patients who died after discharge (4.7% versus 41.7%, p < 0.001) and 8 (3.3%) had new ventricular conduction defects, none of whom died during follow-up.

Conclusions

The presence and extent of repolarization abnormalities predicted outcome in patients with COVID-19. New repolarization abnormalities after discharge were associated with post-discharge mortality.

How safe are children with COVID-19 from cardiac risks? Pediatric risk assesment; insights from echocardiography and electrocardiography

Background/aim

Approximately 40 million individuals worldwide have been infected with SARS-CoV-2, the virus responsible for the novel coronavirus disease-2019 (COVID-19). Despite the current literature about the cardiac effects of COVID-19 in children, more information is required. We aimed to determine both cardiovascular and arrhythmia assessment via electrocardiographic and echocardiographic parameters.

Materials and methods

We evaluated seventy children who were hospitalized with COVID-19 infections and seventy children as normal control group through laboratory findings, electrocardiography (ECG), and transthoracic echocardiography (TTE).

Results

We observed significantly increased levels of Tp-Te, Tp-Te/QT, and Tp-Te/QTc compared with the control group. Twenty-five of 70 (35.7%) patients had fragmented QRS (fQRS) without increased troponin levels. On the other hand, none of the patients had pathologic corrected QT(QTc) prolongation during the illness or its treatment. On TTE, 20 patients had mild mitral insufficiency, among whom only five had systolic dysfunction (ejection fraction < 55%). There was no significant difference between the patient and control groups, except for isovolumic relaxation time (IVRT) in terms of mean systolic and diastolic function parameters. IVRT of COVID patients was significantly lower than that of control group.

Conclusion

Despite all the adult studies, the effects of COVID‐19 on myocardial function are not well established in children. The thought that children are less affected by the illness may be a misconception.

Fragmented QRS is associated with intraventricular dyssynchrony and independently predicts nonresponse to cardiac resynchronization therapy—Systematic review and meta‐analysis

Background

Fragmented QRS (fQRS) is postulated to be associated with ventricular dyssynchrony and might be able to predict a nonresponse to cardiac resynchronization therapy (CRT) implantation. In this systematic review and meta‐analysis, we aim to assess whether fQRS can be a marker of intraventricular dyssynchronies in patients with ischemic and nonischemic cardiomyopathy and whether it is an independent predictor of nonresponse in patients receiving CRT.

Methods

We performed a comprehensive search on topics that assesses fQRS and its association with intraventricular dyssynchrony and nonresponse to CRT up until September 2019.

Results

Fragmented QRS is associated with intraventricular dyssynchrony (OR 10.34 [3.39, 31.54], p < .001; I²: 80% with sensitivity 76.8%, specificity 77%, LR+ 3.3, and LR− 0.3). Subgroup analysis showed that fQRS is associated with intraventricular dyssynchrony in patients with narrow QRS complex (OR 20.92 [12.24, 35.73], p < .001; I²: 0%) and nonischemic cardiomyopathy (OR of 19.97 [12.12, 32.92], p < .001; I²: 0%). Fragmented QRS was also associated with a higher time‐to‐peak myocardial sustained systolic (Ts‐SD) (OR 15.19 [12.58, 17.80], p < .001; I²: 0% and positive Yu index (OR 15.61 [9.07, 26.86], p < .001; I²: 0%). Fragmented QRS has a pooled adjusted OR of OR of 1.70 [1.35, 2.14], p < .001; I²: 62% for association with a nonresponse to CRT. QRS duration is found to be higher in nonresponders group mean difference −8.54 [−13.38, −3.70], p < .001; I²: 70%.

Conclusion

Fragmented QRS is associated with intraventricular dyssynchrony and is independently associated with nonresponse to cardiac resynchronization therapy.

Fragmented QRS - Its significance

Fragment QRS (fQRS) complex is a myocardial conduction abnormality that indicates myocardial scar. It is defined as additional notches in the QRS complex. Though initially fQRS was defined in the setting of normal QRS duration (<120 m s), later it has been expanded to include conditions with wide QRS complexes as in bundle branch block, ventricular ectopy and paced rhythm, when more than 2 notches are present. It is an important, yet often overlooked marker of mortality and arrhythmic events in many cardiac diseases. The significance of fQRS lies in the fact that it just requires a surface ECG for its recording and the value of information about the condition of the heart it dispenses based on the clinical setting. We review the role of fQRS in predicting adverse cardiac events in various conditions.

Measuring agreement among experts in classifying camera images of similar species

Camera trapping and solicitation of wildlife images through citizen science have become common tools in ecological research. Such studies collect many wildlife images for which correct species classification is crucial; even low misclassification rates can result in erroneous estimation of the geographic range or habitat use of a species, potentially hindering conservation or management efforts. However, some species are difficult to tell apart, making species classification challenging-but the literature on classification agreement rates among experts remains sparse. Here, we measure agreement among experts in distinguishing between images of two similar congeneric species, bobcats (Lynx rufus) and Canada lynx (Lynx canadensis). We asked experts to classify the species in selected images to test whether the season, background habitat, time of day, and the visible features of each animal (e.g., face, legs, tail) affected agreement among experts about the species in each image. Overall, experts had moderate agreement (Fleiss' kappa = 0.64), but experts had varying levels of agreement depending on these image characteristics. Most images (71%) had ≥1 expert classification of "unknown," and many images (39%) had some experts classify the image as "bobcat" while others classified it as "lynx." Further, experts were inconsistent even with themselves, changing their classifications of numerous images when they were asked to reclassify the same images months later. These results suggest that classification of images by a single expert is unreliable for similar-looking species. Most of the images did obtain a clear majority classification from the experts, although we emphasize that even majority classifications may be incorrect. We recommend that researchers using wildlife images consult multiple species experts to increase confidence in their image classifications of similar sympatric species. Still, when the presence of a species with similar sympatrics must be conclusive, physical or genetic evidence should be required.

Genotype-phenotype relationship and risk stratification in loss-of-function SCN5A mutation carriers

INTRODUCTION

Loss-of-function (LoF) mutations in the SCN5A gene cause multiple phenotypes including Brugada Syndrome (BrS) and a diffuse cardiac conduction defect. Markers of increased risk for sudden cardiac death (SCD) in LoF SCN5A mutation carriers are ill defined. We hypothesized that late potentials and fragmented QRS would be more prevalent in SCN5A mutation carriers compared to SCN5A-negative BrS patients and evaluated risk markers for SCD in SCN5A mutation carriers.

METHODS

We included all SCN5A loss-of-function mutation carriers and SCN5A-negative BrS patients from our center. A combined arrhythmic endpoint was defined as appropriate ICD shock or SCD.

RESULTS

Late potentials were more prevalent in 79 SCN5A mutation carriers compared to 39 SCN5A-negative BrS patients (66% versus 44%, p = .021), while there was no difference in the prevalence of fragmented QRS. PR interval prolongation was the only parameter that predicted the presence of a SCN5A mutation in BrS (OR 1.08; p < .001). Four SCN5A mutation carriers, of whom three did not have a diagnostic type 1 ECG either spontaneously or after provocation with a sodium channel blocker, reached the combined arrhythmic endpoint during a follow-up of 44 ± 52 months resulting in an annual incidence rate of 1.37%.

CONCLUSION

LP were more frequently observed in SCN5A mutation carriers, while fQRS was not. In SCN5A mutation carriers, the annual incidence rate of SCD was non-negligible, even in the absence of a spontaneous or induced type 1 ECG. Therefore, proper follow-up of SCN5A mutation carriers without Brugada syndrome phenotype is warranted.