ECG Analysis Methods Research Articles

Abstract 311: Machine Learning of the Electrocardiogram to Detect Regional Structural Abnormalities of the Heart

Introduction: Regional dysfunction of cardiac myocardium can result from diverse underlying conditions and cause adverse events even in patients without clinically apparent cardiovascular disease over ejection fraction. Assessment of regional wall motion abnormalities (WMA) currently requires sophisticated imaging that is not feasible for population screening or frequent disease monitoring. Hypothesis: We hypothesized that an AI-enabled ECG model trained on a dataset of qualitative labels could identify and localize wall motion abnormalities with higher accuracy than traditional ECG analysis Methods: In a large academic center, a deep convolutional network was developed with raw 12-lead ECGs and associated reports from 82,424 transthoracic echocardiograms (N = 50,960 patients, age 64 ± 17 years, 54.5% male). 80% of the data was used in model development and the remaining 20% was used for testing. ECGs with ventricular pacing and non-finalized echocardiography reports were excluded. Pre-processing included Kors transformation and beat extraction for input size of 3x300 samples (Figure 1A) and output labels were established by NLP of semi-structured echocardiography reports. Results: ML ECG based models predicted inferior/posterior regional wall motion abnormality with C-statistic of 0.772 (Figure 1B). Physician interpretation of Q-wave ECG pattern was inferior (C-statistic of 0.505). Other segments had similar AUC (apex 0.81, anterior 0.80, lateral 0.75, septal 0.75). Patients with positive ECG screen had +LR 2.70 of having a wall motion abnormality on echocardiography and patients with WMA by echocardiography had decreased survival (LR 0.41, CI: 0.34-0.47, p<0.005) mortality over follow-up of >1500 days (Figure 1C). Conclusion: This study is the first to evaluate the presence and location of regional wall motion abnormalities from inexpensive and noninvasive 12-lead ECG. This approach could be applied for widespread ambulatory screening.

Statistical method for ECG analysis and diagnostic

In this paper, a statistical method of ECG analysis and diagnostic is proposed. This method is based on three parts: data simplification using multiscaled PCA, faults detection and localisation by introducing classic linear PCA. The studied data is presented as a multivariate matrix. The variables of this matrix are extracted from the ECG waves characteristics: waves amplitudes and segments measurements. The developed approach allows detecting arrhythmias and heart beat troubles. Comparing the results obtained by this approach and the data of the expert, we approve the performance of our study.

Statistical method for ECG analysis and diagnostic

In this paper, a statistical method of ECG analysis and diagnostic is proposed. This method is based on three parts: data simplification using multiscaled PCA, faults detection and localisation by introducing classic linear PCA. The studied data is presented as a multivariate matrix. The variables of this matrix are extracted from the ECG waves characteristics: waves amplitudes and segments measurements. The developed approach allows detecting arrhythmias and heart beat troubles. Comparing the results obtained by this approach and the data of the expert, we approve the performance of our study.

Evaluating a hierarchical approach for heartbeat classification from ECG

Several types of arrhythmias that can be rare and harmless, but may result in serious cardiac issues, and several ECG analysis methods have been proposed in the literature to automatically classify the various classes of arrhythmias. Following the Association for the Advancement of Medical Instrumentation (AAMI) standard, 15 classes of heartbeats can be hierarchically grouped into five superclasses. In this work, we propose to employ the hierarchical classification paradigm to five ECG analysis methods in the literature, and compare their performance with flat classification paradigm. In our experiments, we use the MIT-BIH Arrhythmia Database and analyse the use of the hierarchical classification following AAMI standard and a well-known and established evaluation protocol using five superclasses. The experimental results showed that the hierarchical classification provided the highest gross accuracy for most of the methods used in this work and provided an improvement in classification performance of N and SVEB superclasses.

Evaluating a hierarchical approach for heartbeat classification from ECG

Several types of arrhythmias that can be rare and harmless, but may result in serious cardiac issues, and several ECG analysis methods have been proposed in the literature to automatically classify the various classes of arrhythmias. Following the Association for the Advancement of Medical Instrumentation (AAMI) standard, 15 classes of heartbeats can be hierarchically grouped into five superclasses. In this work, we propose to employ the hierarchical classification paradigm to five ECG analysis methods in the literature, and compare their performance with flat classification paradigm. In our experiments, we use the MIT-BIH Arrhythmia Database and analyse the use of the hierarchical classification following AAMI standard and a well-known and established evaluation protocol using five superclasses. The experimental results showed that the hierarchical classification provided the highest gross accuracy for most of the methods used in this work and provided an improvement in classification performance of N and SVEB superclasses.

DYNAMIC 3-DIMENSIONAL ECG ASSESSMENT IN INDIVIDUALS WITH LONG-QT SYNDROME AND ARVC AND ABORTED CARDIAC ARREST

Ventricular arrhythmias related to inherited arrhythmias are triggered by repolarization or depolarization abnormalities resulting from abnormal ion channel function. Non-invasive markers of abnormal repolarization/depolarization are useful for risk prediction of ventricular arrhythmia. However, conventional ECG markers have well-recognized limitations. Three-dimensional vectorcardiography is a promising novel method of ECG analysis, but little is known about its potential role in manifest inherited arrhythmias.

AR identification and spectral estimate applied to the R-R interval measurements

The methods of identification and spectral estimate are applied to the tachogram, i.e. the time series constituted by the cycle-by-cycle R-R interval durations measured on the ECG signal from cardiological patients in ambulatory rehabilitation training after episodes of myocardial infarction or ischemic disease. The Batch Least Squares Method is applied to identify the series as an AR process of 5th order. The whiteness test and Rissanen's optimization criterion are also fulfilled. The clinical information is in this way highly compressed in the pole diagram and in the Maximum Entropy Spectrum (MES) estimated on the basis of the AR coefficients. The experimental results in a restricted set of patients confirm the feasibility of new instrumentation design criteria for non-conventional R-R intervals parametrisation, successive diagnostic classification and beat prediction. Finally, some preliminary considerations about the capabilities of the introduced methods put into evidence the role of computerized techniques in recognizing the fundamental patterns of physiopathological heart rate variability, which the usual conventional methods of ECG analysis are not able to detect in a reliable way.

Status Report on Computerized ECG Analysis

COMPUTER analysis of ECGs has been evolving during the past 15 years. In 1976, an estimated 80 million ECGs were performed in the United States, and approximately 4 million of these were analyzed by computers. 1 In the past two years, three factors have combined to accelerate the rate of utilization of this method: the clinical diagnostic accuracy of some programs has increased to a level at which some clinicians have judged the programs to be useful to them in their care of patients; comprehensive computer systems have been made available that allow users to perform editing of interpretations, organization of reports, and billing; and the price-performance ratio of computer equipment has decreased substantially. Diagnostic Accuracy The most important variable in the evaluation of the potential benefits of this method of ECG analysis is the diagnostic accuracy of the computer programs. Unfortunately, this variable is very difficult to measure, and

Status report on computerized ECG analysis

COMPUTER analysis of ECGs has been evolving during the past 15 years. In 1976, an estimated 80 million ECGs were performed in the United States, and approximately 4 million of these were analyzed by computers.¹In the past two years, three factors have combined to accelerate the rate of utilization of this method: the clinical diagnostic accuracy of some programs has increased to a level at which some clinicians have judged the programs to be useful to them in their care of patients; comprehensive computer systems have been made available that allow users to perform editing of interpretations, organization of reports, and billing; and the price-performance ratio of computer equipment has decreased substantially. <h3>Diagnostic Accuracy</h3> The most important variable in the evaluation of the potential benefits of this method of ECG analysis is the diagnostic accuracy of the computer programs. Unfortunately, this variable is very difficult to measure, and

A test experience with a machine-processed electrocardiography diagnosis: The recognition of “normal” and some specific patterns

Five hundred ECG's, machine processed by the commercially available IBM Bonner-Mt. Sinai System, have been reviewed. The diagnostic printouts were of value as an adjunct to the physician's interpretations, but were not sufficiently reliable--in regard to either sensitivity (identification of the abnormal or borderline) or specificity (accuracy in predicting typical abnormalities)--to allow us to condone any claim for it being an independent reliable diagnostic service to noncardiologist clinicians. Despite this desclaimer for current diagnostic reliability of ECG computer analyses, specifically the IBM-Bonner system, we continue to be of the firm conviction that progress through this method of ECG analysis will have much to contribute in future years.

Computer classifications of electrocardiograms

Several statistical classification techniques were applied to orthogonal electrocardiographic record samples, obtained from normal subjects and patients with left ventricular hypertrophy. A simple method based on addition of QRS amplitude measurements was used as representative of ECG analysis methods in present clinical use. The procedures to be evaluated and compared consisted of vector differences with and without weight factors, and a class-separating and a class-clustering transformation. They were tested both with sets of three and eight amplitude measurements of the QRS complex by using different record samples of 100 each. Best results were obtained with weighted vector differences based on eight amplitudes (84% correct classifications). The class-separating procedure followed with 80%. The conventional method of summing amplitudes led to 67% separation. When amplitude measurements were decreased from eight to three, the differentiation of records deteriorated by 5–10%. From the results it was concluded that improvement of ECG record classification can be achieved mainly through increase in number of measurements. More complex statistical classification methods lead only to modest improvements with small numbers of measurements but to a substantial enhancement when more measurements become available. These results indicate a need for automatic means for ECG data analysis because larger numbers of ECG measurements and more efficient classification methods are not practical without access to computer facilities.