A patient adaptable ECG beat classifier based on neural networks
Introduction
The electrocardiogram (ECG) is a graphic recording of the electrical activities in human heart and provides diagnostically significant information. Its shape, size and duration reflect the heart rhythm over time. The waves related to electrical impulses occurring at each beat of the heart are shown in Fig. 1. The P-wave represents the beginning of the cardiac cycle and is followed by the QRS complex, which is generally the most recognizable feature of an ECG waveform. At the end of the cardiac cycle is the T-wave. The varied sources of heart diseases provide a wide range of alterations in the shape of the electrocardiogram. For instance, inverted T waves (Fig. 2) are seen during the evolution of myocardial infarction, while ST-segment depression (Fig. 3) can be caused by ischemia. In recent years, many researches concerning automated processing of ECG signals have been conducted ([1], [4], [5], [7], [10], [11], [13], [14]). A difficult problem in computer-aided ECG analysis is related to the large variation in the morphology of ECG waveforms, not only among different patients, but even within the same patient. This makes the detection of ECG features (ST-segment, T-wave, QRS-area) a tough task.
The aim of the present work is to design an ECG beat recognition method to distinguish normal and ischemic patterns of the same patient without requiring the extraction of ECG features. The method is a supervised neural network-based algorithm, and hence its applicability requires the availability of recordings of both normal beats and annotated ischemic episodes. Once trained, the system should be capable of detecting new ischemic beats similar to those previously observed in the patient. This objective is relevant because of the observed tendency of patients discharged after an acute myocardial ischemic attack to repeat a similar ischemic episode within a short period of time (up to 30% within one month). The possibility of automatically monitoring such patients would then translate directly into reduced morbidity and mortality.
In the proposed method, the pattern under classification is defined to be the ECG signal from an RRR interval, i.e. the ECG signal from two successive heart beats, marked by three successive R peaks. We remark that RRR intervals can be easily detected and contain enough information for pattern recognition. The patterns of the RRR intervals can be represented as vectors lying in relatively high dimensional spaces. However, due to the variation of the ECG waveform, to variations in heart rate and to the presence of noise, vectors may be very different from each other and actually lie in different dimensional spaces. This makes it difficult to train a neural network for beat recognition. The proposed strategy is that of considering the ECG data from an RRR interval as a noisy sample of an underlying function. The function is approximated by means of a linear combination of a fixed number of suitable basis functions. The coefficients of the linear expansion can be easily computed by solving a linear least-squares problem, and constitute the extracted features of the RRR interval. The transformed patterns, i.e. the coefficients of the linear expansions, become the input signals of a feed-forward neural network classifier, which provides an output of zero for the normal and one for the ischemic case. The adopted technique merges all the patterns to be classified in the same lower dimensional space and reduces the influence of ECG morphology and noise.
In Section 2 we describe the approximation technique of RRR interval and the neural network ECG beat classifier. In Section 3 we present the results of the tests that were carried out in order to evaluate the designed system as beat classifier. Finally, Section 4 contains some concluding remarks.
Section snippets
Approximation of RRR interval and beat classification using neural networks
The pattern under classification is the ECG digital recording of two consecutive RR-segments (hereafter defined RRR interval). Note that
- (i)
R peaks can be easily detected in the digitized signal, since they are local maxima that can be exactly determined once known the period of the heart cycle and the sampling frequency.
- (ii)
The RRR interval contains two complete heart cycles and hence it constitutes suitably informative input of a pattern recognition system.
- (iii)
Sampled ECG over an RRR interval is a vector
Experimental results and discussion
The described pattern recognition system is targeted to a single patient, in the sense that it is trained using the ECG of that patient in order to correctly distinguish his/her normal and ischemic beats. The European ST-T Database [6] was used as source of experimental ECG’s. Among the records having dominant R peaks (the method requires extraction of RRR intervals), we randomly selected a subset of 20 records. For each considered record (i.e. for each considered patient), a set of 400 RRR
Conclusion
In this work we have presented a supervised neural network-based algorithm for ECG ischemic beat recognition of the same patient. The obtained results show that the proposed algorithm offers a good combination of sensitivity and specificity, making the design of a practical automatic ischemia detector feasible. The proposed approach leads to a dedicated model for each patient, so that, its applicability requires the availability of recordings of annotated ischemic episodes. Therefore, it could
References (16)
- et al.
Feature measurements of ECG beats based on statistical classifiers
Measurement
(2007) - et al.
An ischemia detection method based on artificial neural networks
Artificial Intelligence in Medicine
(2002) - et al.
Detection of ECG Arrhythmia using a differential expert system approach based on principal component analysis and least square support vector machine
Applied Mathematics and Computation
(2007) - et al.
Classification of cardiac abnormalities using heart rate signals
Medical and Biological Engineering and Computing
(2004) Neural Networks for Pattern Recognition
(1995)Numerical Methods for Least Squares Problems
(1996)- et al.
An association rule mining based methodology for the automated detection if ischemic ECG beats
IEEE Transactions on Biomedical Engineering
(2006) - European Society of cardiology, European ST-T database directory, S.T.A.R., Pisa,...
Cited by (40)
Short and noisy electrocardiogram classification based on deep learning
2020, Deep Learning for Data Analytics: Foundations, Biomedical Applications, and ChallengesHybrid model based on neural networks, type-1 and type-2 fuzzy systems for 2-lead cardiac arrhythmia classification
2019, Expert Systems with ApplicationsCitation Excerpt :Different methods have been put forward in current literature to help in the interpretation of the different morphologies or electrocardiographic patterns that represent different diseases that are diagnosed by interpreting the electrocardiograms (Luis, Nunes, & Alburquerque, 2013; Luz, Schwartz, Camara, & Menotti, 2015; Merone, Soda, & Sansone, 2017). Such as learning vector quantization (LVQ) (Amezcua & Melin, 2014; Melin, Amezcua, Valdez, & Castillo, 2014), multilayer perceptron (MLP) (Gaetano, Panunzi, Risi, & Sciandrone, 2009; Javadi, Asgnar, Sajedin, & Ebrahimpour, 2013; Jovic & Bogunovic, 2012; Rahhal et al., 2016; Ramirez, Castillo, & Soria, 2010; Megat, Jahidin & Norall, 2012; Castillo, Melin, Ramirez, & Soria, 2012; Ceylan, Ozbay, & Karlik, 2009; Ozbay & Tezel, 2010; Yu & Chou, 2008), Markov models, self organizing maps (SOM), linear discriminant systems, Bayesian approach, support vector machines (SVM) (Kalaf, Owis & Yassine, 2015; Homaeinezhad et al., 2012; Moavenian & Khorrami, 2010), higher order statistics (Martis, Achayra, & Min, 2013; Martis, Achayra, Prasad, & Chua, 2013), expert systems, type-1 fuzzy systems (Castillo et al., 2012; Ramirez et al., 2010; Wang, Chiang, Hsu, & Yang, 2013), type-2 fuzzy systems (Chua & Tan, 2009; Melin & Castillo, 2013; Tan, FOO, & Chua, 2007), statistical and syntactic pattern recognition, fuzzy KNN algorithm (Castillo et al., 2012; Keller, Gray, & Givens, 1985; Ramirez et al., 2010), as well as hybrid systems (Gacek & Witold, 2012; Melin & Castillo, 2013; Melin, Prado, Miramontes, & Medina, 2016; Melin, Ramirez, & Prado, 2017; Osowki, Markiewicz, & Hoal, 2018; Osowksi, Siwek, & Siroic, 2011; Wozniak, Grana, & Corchado, 2014; Jadhav, Nalbalwar & Ghatol, 2010; Nazmy, EL-Messiry, & AL-Bokhity, 2010; Ozbay & Tezel, 2010; Ozbay, Ceylan, & Karlik, 2005; Shao, Hou, & Chiu, 2014; Zopounidis & Doumpos, 2002), Other solutions implement optimization algorithms such as Ant Colony and Particle Swarm Optimization (Edward, Ramu, & Swaminathan, 2016; Koruek & Dogan, 2010) . Some solutions are focus on the preprocessing process such as PCA, LDA, ICA, and Discrete Wavelet Transform (Martis, Acharya, Mandana, Ray, & Chakraborty, 2012), linear and nonlinear features (Elhaj, Salim, Harris, Swee, & Ahmed, 2016; Sufi, Khalil & Naser, 2011; Nien, & Chou, 2009).
A simple realtime algorithm for automatic external defibrillator
2019, Biomedical Signal Processing and ControlModeling ischemia with finite elements and automated machine learning
2018, Journal of Computational ScienceCitation Excerpt :The same research group created an expert system for classification of ischemic beats based on decision trees [11]. De Gaetano et al. [12] adapted neural networks to create a patient adaptable classifier for prediction of ischemic ECG beats. Ansari et al. [13] provided a recent review of a large set of prediction approaches, focusing on automated detection of ischemia using ECG and electronic health records.
A survey on ECG analysis
2018, Biomedical Signal Processing and ControlMathematical model with autoregressive process for electrocardiogram signals
2018, Communications in Nonlinear Science and Numerical Simulation