Frank vectorcardiographic system from standard 12 lead ECG: An effort to enhance cardiovascular diagnosis

AI-enhanced reconstruction of the 12-lead electrocardiogram via 3-leads with accurate clinical assessment

The 12-lead electrocardiogram (ECG) is an integral component to the diagnosis of a multitude of cardiovascular conditions. It is performed using a complex set of skin surface electrodes, limiting its use outside traditional clinical settings. We developed an artificial intelligence algorithm, trained over 600,000 clinically acquired ECGs, to explore whether fewer leads as input are sufficient to reconstruct a 12-lead ECG. Two limb leads (I and II) and one precordial lead (V3) were required to generate a reconstructed 12-lead ECG highly correlated with the original ECG. An automatic algorithm for detection of ECG features consistent with acute myocardial infarction (MI) performed similarly for original and reconstructed ECGs (AUC = 0.95). When interpreted by cardiologists, reconstructed ECGs achieved an accuracy of 81.4 ± 5.0% in identifying ECG features of ST-segment elevation MI, comparable with the original 12-lead ECGs (accuracy 84.6 ± 4.6%). These results will impact development efforts to innovate ECG acquisition methods with simplified tools in non-specialized settings.

Intelligent assessment of atrial fibrillation gradation based on sinus rhythm electrocardiogram and baseline information

BACKGROUND

Atrial fibrillation (AF) is a progressive arrhythmia that significantly affects a patient's quality of life. The 4S-AF scheme is clinically recommended for AF management; however, the evaluation process is complex and time-consuming. This renders its promotion in primary medical institutions challenging. This retrospective study aimed to simplify the evaluation process and present an objective assessment model for AF gradation.

METHODS

In total, 189 12-lead electrocardiogram (ECG) recordings from 64 patients were included in this study. The data were annotated into two groups (mild and severe) according to the 4S-AF scheme. Using a preprocessed ECG during the sinus rhythm (SR), we obtained a synthesized vectorcardiogram (VCG). Subsequently, various features were calculated from both signals, and age, sex, and medical history were included as baseline characteristics. Different machine learning models, including support vector machines, random forests (RF), and logistic regression, were finally tested with a combination of feature selection techniques.

RESULTS

The proposed method demonstrated excellent performance in the classification of AF gradation. With an optimized feature set of VCG and baseline features, the RF model achieved accuracy, sensitivity, and specificity of 83.02 %, 80.56 %, and 88.24 %, respectively, under the inter-patient paradigm.

CONCLUSION

Our results demonstrate the value of physiological signals in AF gradation evaluation, and VCG signals were effective in identifying mild and severe AF. Considering its low computational complexity and high assessment performance, the proposed model is expected to serve as a useful prognostic tool for clinical AF management.

AI-Enhanced Reconstruction of the 12-Lead Electrocardiogram via 3-Leads with Accurate Clinical Assessment

The 12-lead electrocardiogram (ECG) is an integral component to the diagnosis of a multitude of cardiovascular conditions. It is performed using a complex set of skin surface electrodes, limiting its use outside traditional clinical settings. We developed an artificial intelligence algorithm, trained over 600,000 clinically acquired ECGs, to explore whether fewer leads as input are sufficient to reconstruct a full 12-lead ECG. Two limb leads (I and II) and one precordial lead (V3) were required to generate a reconstructed synthetic 12-lead ECG highly correlated with the original ECG. An automatic algorithm for detection of acute myocardial infarction (MI) performed similarly for original and reconstructed ECGs (AUC=0.94). When interpreted by cardiologists, reconstructed ECGs achieved an accuracy of 81.4±5.0% in identifying ST elevation MI, comparable with the original 12-lead ECGs (accuracy 84.6±4.6%). These results will impact development efforts to innovate ECG acquisition methods with simplified tools in non-specialized settings.

Personalized LSTM Models for ECG Lead Transformations Led to Fewer Diagnostic Errors Than Generalized Models: Deriving 12-Lead ECG from Lead II, V2, and V6

BACKGROUND AND OBJECTIVE

The prevalence of chronic cardiovascular diseases (CVDs) has risen globally, nearly doubling from 1990 to 2019. ECG is a simple, non-invasive measurement that can help identify CVDs at an early and treatable stage. A multi-lead ECG, up to 15 leads in a wearable form factor, is desirable. We seek to derive multiple ECG leads from a select subset of leads so that the number of electrodes can be reduced in line with a patient-friendly wearable device. We further compare personalized derivations to generalized derivations.

METHODS

Long-Short Term Memory (LSTM) networks using Lead II, V2, and V6 as input are trained to obtain generalized models using Bayesian Optimization for hyperparameter tuning for all patients and personalized models for each patient by applying transfer learning to the generalized models. We compare quantitatively using error metrics Root Mean Square Error (RMSE), R², and Pearson correlation (ρ). We compare qualitatively by matching ECG interpretations of board-certified cardiologists.

RESULTS

ECG interpretations from personalized models, when corrected for an intra-observer variance, were identical to the original ECGs, whereas generalized models led to errors. Mean performance values for generalized and personalized models were (RMSE-74.31 µV, R²-72.05, ρ-0.88) and (RMSE-26.27 µV, R²-96.38, ρ-0.98), respectively.

CONCLUSIONS

Diagnostic accuracy based on derived ECG is the most critical validation of ECG derivation methods. Personalized transformation should be sought to derive ECGs. Performing a personalized calibration step to wearable ECG systems and LSTM networks could yield ambulatory 15-lead ECGs with accuracy comparable to clinical ECGs.

Deep-Learning-Based Estimation of the Spatial QRS-T Angle from Reduced-Lead ECGs

The spatial QRS-T angle is a promising health indicator for risk stratification of sudden cardiac death (SCD). Thus far, the angle is estimated solely from 12-lead electrocardiogram (ECG) systems uncomfortable for ambulatory monitoring. Methods to estimate QRS-T angles from reduced-lead ECGs registered with consumer healthcare devices would, therefore, facilitate ambulatory monitoring. (1) Objective: Develop a method to estimate spatial QRS-T angles from reduced-lead ECGs. (2) Approach: We designed a deep learning model to locate the QRS and T wave vectors necessary for computing the QRS-T angle. We implemented an original loss function to guide the model in the 3D space to search for each vector's coordinates. A gradual reduction of ECG leads from the largest publicly available dataset of clinical 12-lead ECG recordings (PTB-XL) is used for training and validation. (3) Results: The spatial QRS-T angle can be estimated from leads {I, II, aVF, V2} with sufficient accuracy (absolute mean and median errors of 11.4° and 7.3°) for detecting abnormal angles without sacrificing patient comfortability. (4) Significance: Our model could enable ambulatory monitoring of spatial QRS-T angles using patch- or textile-based ECG devices. Populations at risk of SCD, like chronic cardiac and kidney disease patients, might benefit from this technology.

QRS area as a predictor of cardiac resynchronization therapy response: a systematic review and meta-analysis

BACKGROUND

QRS area, a three-dimensional QRS complex, is a novel vectorcardiography method of measuring the magnitude of electrical forces in the heart. Hypothetically, a greater QRS area denotes higher dyssynchrony and indicates potential benefits from cardiac resynchronization therapy (CRT). Previous studies suggest a positive correlation between QRS area and the degree of response to CRT, but its clinical use remains unclear. We performed a meta-analysis of the relationship between QRS area and survival benefit following CRT.

METHODS

We comprehensively searched the MEDLINE, EMBASE, and Cochrane databases from inception to August 2021. We included studies with prospective and retrospective cohort designs that reported QRS area before CRT and total mortality. Data from each study were analyzed using a random-effects model. The results were reported as a hazard ratio (HR) and 95% confidence intervals.

RESULTS

Five observational studies including 4,931 patients were identified. The cut-off values between large and small QRS areas ranged from 102-116 μVs. Our analysis showed a larger QRS area was statistically associated with increased 5-year survival in patients implanted with CRT (HR pooled 0.48, 95% CI 0.46-0.51, I² = 54%, P < 0.0001). Greater QRS area reduction (pre- and post-implantation) were associated with a lower total mortality rate (HR pooled 0.45, 95% CI 0.38-0.52, I² = 0%, P < 0.0001).

CONCLUSION

Larger pre-implantation QRS area was associated with increased survival after CRT. QRS area reduction following CRT implantation was also associated with lower mortality. QRS area may potentially become an additional selection criterion for CRT implantations. This article is protected by copyright. All rights reserved.

Automatic Classification of Myocardial Infarction Using Spline Representation of Single-Lead Derived Vectorcardiography

Myocardial infarction (MI) is one of the most prevalent cardiovascular diseases worldwide and most patients suffer from MI without awareness. Therefore, early diagnosis and timely treatment are crucial to guarantee the life safety of MI patients. Most wearable monitoring devices only provide single-lead electrocardiography (ECG), which represents a major limitation for their applicability in diagnosis of MI. Incorporating the derived vectorcardiography (VCG) techniques can help monitor the three-dimensional electrical activities of human hearts. This study presents a patient-specific reconstruction method based on long short-term memory (LSTM) network to exploit both intra- and inter-lead correlations of ECG signals. MI-induced changes in the morphological and temporal wave features are extracted from the derived VCG using spline approximation. After the feature extraction, a classifier based on multilayer perceptron network is used for MI classification. Experiments on PTB diagnostic database demonstrate that the proposed system achieved satisfactory performance to differentiating MI patients from healthy subjects and to localizing the infarcted area.

Comparison of spatial temporal representations of the vectorcardiogram using digital image processing

INTRODUCTION

The vectorcardiography (VCG) is a method of representing the heart's electrical activity in three dimensions that is not frequently used in clinical practice due to the higher complexity compared to electrocardiography (ECG). A way around this problem was the development of regression techniques to obtain the VCG from the 12‑lead ECG and the evaluation of these techniques is done by comparing the parameters obtained by the gold standard method and by the VCG obtained by the alternative methods. In this paper it is proposed instead a comparison between the images of the VCG planes using the values returned by digital image processing metrics such as PSNR, SSIM and PW-SSIM.

METHODS

The signals used were obtained from the Physikalisch-Technische Bundesanstalt Diagnostic ECG Database, which contains both the VCGs obtained by the gold standard method and the 12 lead ECG signals. They were divided into five groups that contained a control group and according to the region of the wall infarction. The ECG signals were then filtered using a Butterworth Finite Impulse Response bandpass filter, with cutoff frequencies of 3 Hz and 45 Hz and then the VCGs were by a computer application using the Kors inverse matrix method, the Kors quasi-orthogonal method and the Dower Inverse Matrix method. The reconstructed signals were then compared using the PSNR, SSIM and PW-SSIM methods. The returned values were presented in tables for each group containing the average value and standard deviance for each method in each VCG plane.

RESULTS

Using image processing techniques, it was possible to perceive that the alternative methods to obtain the VCG have a high confiability that could be compared to the gold standard in signals from healthy subjects. However, signals from pathological subjects present variations that could be caused by a deficit of these alternative methods to represent the pathology in these cases. Considering the PW-SSIM, the Frontal plane by the reconstructions was considered the most similar to the gold standard, having PW-SSIM values higher than 0.93 and for the Horizontal plane two groups had PW-SSIM values lower than 0.90 and for the Sagittal plane all groups had values lower than this value.

DISCUSSION

The values yielded by the PSNR and SSIM had low variance, worsening the perception of the effect of the reconstruction method used or the infarction effect over the reconstruction. The values lower than 0.90 could indicate that these planes have their generation most affected by the infarction.

CONCLUSION

The three methods of obtaining the VCG Frank leads, the Kors Quasi-Orthogonal method, the Kors Linear Regression method and the Dower Inverse Matrix, presented differences in the metrics: PSNR, SSIM and PW-SSIM in normal subjects according to the planes frontal, horizontal and sagittal and in subjects with Myocardial Infarction according to its topography: anterior, inferolateral, inferior or multiarterials. Considering only the PW-SSIM, the QO method had the best performance in different signals, followed by the Dower method.

Heterogeneous Recurrence Analysis of Disease-Altered Spatiotemporal Patterns in Multi-Channel Cardiac Signals

Heart diseases alter the rhythmic behaviors of cardiac electrical activity. Recent advances in sensing technology bring the ease to acquire space-time electrical activity of the heart such as vectorcardiogram (VCG) signals. Recurrence analysis of successive heartbeats is conducive to detect the disease-altered cardiac activities. However, conventional recurrence analysis is more concerned about homogeneous recurrences, and overlook heterogeneous types of recurrence variations in VCG signals (i.e., in terms of state properties and transition dynamics). This paper presents a new framework of heterogeneous recurrence analysis for the characterization and modeling of disease-altered spatiotemporal patterns in multi-channel cardiac signals. Experimental results show that the proposed approach yields an accuracy of 96.9%, a sensitivity of 95.0%, and a specificity of 98.7% for the identification of myocardial infarctions. The proposed method of heterogeneous recurrence analysis shows strong potential to be further extended for the analysis of other physiological signals such as electroencephalogram (EEG) and electromyography (EMG) signals towards medical decision making.

Plug and play self-configurable IoT gateway node for telemonitoring of ECG

In the era of IoT and hyperconnection, an efficient electrocardiogram (ECG) telemonitoring system in wireless body area network (WBAN) demands an easy to use, self-configurable, secure, plug and play system with minimum hardware and computational complexities. The compression and quantization parameters required for an efficient representation of ECG signal will vary from patient to patient, from lead to lead, and from time to time. To this end, we propose a compressed sensing based WBAN with self-configurable gateway node (CS-SCGN) using deterministic binary block diagonal (DBBD) measurement matrix. The self-configurability is brought in through a low complex method for adaptive tuning of parameters with a careful choice of measurement matrix and data length. The redundant data transfer between sensor nodes and gateway node is avoided by addressing the diverse requirements in ECG remote health monitoring through three modes of configuration in the proposed system. A further reduction in communication and storage cost is achieved by optimizing the number of bits transmitted by sensor nodes by automatically tuning the compression ratio and quantization depth based on the dynamics of ECG signal. The self-configuration algorithm is designed to run at the gateway node in such a way as to optimize the power efficiency of sensor nodes without causing an extra power drain at the gateway node. Also, we investigate the feasibility of using smartphone as an IoT gateway node for performing primary processing to provide local utility before sending the received data to the remote server. The energy efficiency and real-time feasibility of the proposed algorithm are evaluated by implementing the gateway node on Odroid-XU4 board which runs on the same processor as in the latest smartphones. The experimental results indicate that our proposed self-configurable system at the gateway node makes the entire ECG telemonitoring system flexible, plug and play, patient independent and power-efficient.

Comparison of Different Electrocardiography with Vectorcardiography Transformations

This paper deals with transformations from electrocardiographic (ECG) to vectorcardiographic (VCG) leads. VCG provides better sensitivity, for example for the detection of myocardial infarction, ischemia, and hypertrophy. However, in clinical practice, measurement of VCG is not usually used because it requires additional electrodes placed on the patient's body. Instead, mathematical transformations are used for deriving VCG from 12-leads ECG. In this work, Kors quasi-orthogonal transformation, inverse Dower transformation, Kors regression transformation, and linear regression-based transformations for deriving P wave (PLSV) and QRS complex (QLSV) are implemented and compared. These transformation methods were not yet compared before, so we have selected them for this paper. Transformation methods were compared for the data from the Physikalisch-Technische Bundesanstalt (PTB) database and their accuracy was evaluated using a mean squared error (MSE) and a correlation coefficient (R) between the derived and directly measured Frank's leads. Based on the statistical analysis, Kors regression transformation was significantly more accurate for the derivation of the X and Y leads than the others. For the Z lead, there were no statistically significant differences in the medians between Kors regression transformation and the PLSV and QLSV methods. This paper thoroughly compared multiple VCG transformation methods to conventional VCG Frank's orthogonal lead system, used in clinical practice.

T wave axis deviation and QRS-T angle - Controversial indicators of incident coronary heart events

Abnormal orientation of the T-wave axis and increased angle between the QRS complex (depolarization) and the T-wave (repolarization) have long been assumed to provide a global measure of repolarization abnormality, and have been used to assess ventricular repolarization. The ability of the T wave axis deviation and the QRS-T angle to predict incident coronary heart events was examined in several studies. However, conflicting results have led to significant controversy in the literature concerning their purported ability. Potential explanations involve true variation between study populations, non-standardized cut-off values, different baseline cardiovascular risk levels or different patterns of confounding by other concomitant cardiovascular risk factors. In the present article we will attempt to briefly present the rationale and pathophysiology behind these indices, summarize existing knowledge regarding their prognostic significance and their correlation with established cardiovascular disease risk factors. Further prospective studies are necessary to confirm or refute whether T-wave axis deviation, QRS-T angle and ventricular gradient may in the future serve as indicators of incident coronary heart events and mortality, both in populations with higher prevalence of subclinical advanced atherosclerotic heart disease and in apparently healthy subjects.

Professor André Jouve (1909-2001): A French cardiologist who was a pioneer of clinical vector-electrocardiography and cardiovascular epidemiology

André, Julien, Auguste Jouve was born in Marseilles on June 10, 1909 son of Xavier Marie Francois Louis Jouve MD and Marie Louise Charlotte Vigliengo his wife. He had a brilliant medical career in Marseilles: Resident at Marseilles Hospitals in 1931, major of his promotion, then an Assistant in 1943 and a Chief in 1951, to become Associate Professor of Medicine in 1946 and finally Full Professor of Clinical and Experimental Cardiology in 1954. Fellow of several Cardiological Societies, he became President of the French Society of Cardiology in 1968, Vice-President of the European Society of Cardiology in 1972 and finally President of the French College of Vascular Pathology in 1973. He had been a WHO Expert for degenerative and cardiovascular diseases from 1958 to 1981 and a National correspondent of the Academy of Medicine in 1977. He was decorated by the Légion d'Honneur (Officer in 1975). He retired in 1981 and died in 2001. Clinical vector-electrocardiology and cardiovascular epidemiology were the main areas of his interest where he made essential contributions such as the famous treatise on ECG. The Heart Cantini Center was considered his leading creation and action, where the first French heart transplantation was performed in 1968, the first French epidemiological investigation on coronary risk factors took place, the idea of starting prevention at pediatric age was clearly outlined and the need of concentrating on psychological and dysmetabolic factors was precisely advocated for indexing later development of ischemic heart diseases. These achievements are reviewed and put into perspective.