No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Detection of Transient ST Segment Episodes During Ambulatory ECG Monitoring
Abstract
Using the European Society of Cardiology ST-T Database, we have developed a Karhunen-Loève transform-based algorithm for robust automated detection of transient ST segment episodes during ambulatory ECG monitoring. We review current approaches and systems to detect transient ST segment changes and describe the architecture of our algorithm and its development. The algorithm incorporates a single-scan trajectory-recognition technique in feature space using the Mahalanobis distance function between the feature vectors. The main characteristics of the algorithm are detection of noisy beats, correction of the reference ST segment level to correct for slow ST level drift, detection of sudden significant shifts of ST deviation due to shifts of the mean electrical axis of the heart, detection of transient ST episodes, and, by tracking the QRS complex morphology, differentiation between ischemic and nonischemic ST episodes as a result of axis shifts. We compared the algorithm's performance to other recently developed algorithms and estimated its real-world performance.
... It then fits a third order cubic spline polynomial to these knots to obtain the baseline estimate which is then subtracted from the original signal. This method is used in a large number of MI and ischemic beat detection algorithms [11,22,23,24,30,36,37,65,66,67,68,69,70,71,72,73,74]. However, the performance of this method is highly dependent on the accuracy of the PQ segment detection algorithm. ...
... For example, Minchole et al. [71] and García et al. [36] used a linear phase finite impulse response (FIR) filter with a cutoff frequency of 25Hz. Moreover, Jager et al. [11,69] used a 6 th order Butterworth low-pass filter with a cutoff frequency of 55Hz while Safdarian et al. [88] and Kora and Kalva [89] used Butterworth and Sgolay filters, respectively, without specifying the frequency and order of the filters. Low-pass filtering can also be conducted using moving average filters [61,88]. ...
... In [98], the abnormal beats and their neighboring beats were discarded when the difference between the PQ level and the mean PQ level was larger than 0.6mV. In [69], noisy beats were detected using an algorithm proposed in [104] which thresholds the peak-topeak amplitude of the ECG beat. In [23,24,30], the authors excluded the noisy beats whose RMS noise level, measured within a 40ms window located at two thirds of the RR interval, is larger than 0.04mV. ...
There is a growing body of research focusing on automatic detection of ischemia and myocardial infarction (MI) using computer algorithms. In clinical settings, ischemia and MI are diagnosed using ECG recordings as well as medical context including patient symptoms, medical history and risk factors, information that is often stored in the electronic health records. The ECG signal is inspected to identify changes in the morphology such as ST-segment deviation and T wave changes. Some of the proposed methods compute similar features automatically while others use non-conventional features such as wavelet coefficients. This paper provides an overview of the methods that have been proposed in this area, focusing on their historical evolution, the publicly available datasets that they have used to evaluate their performance and the details of their algorithms for ECG and EHR analysis. The validation strategies that have been used to evaluate the performance of the proposed methods are also presented. Finally, the paper provides recommendations for future research to address the shortcomings of the currently existing methods and practical considerations to make the proposed technical solutions applicable in clinical practice.
... An increased heart rate and transient morphology change of the ST segments of heart beats may be observed. The lower data segment (Fig 1B) is an example of severe noise which is often present in AECG records and cause the main problems during the visual and automatic assessing of the severity of ischaemic ST segment episodes. of using the KLT in ECG signal analysis was noise estimation [4], visually identifying acute ischaemic episodes [5], the representation of ECG morphology [4,6], the automated detection of transient ST segment episodes during AECG monitoring [7,8], the analysis of the cardiac repolarization period (ST-T complex) [9][10][11][12], visually identifying and manually annotating the transient ischaemic and non-ischaemic ST segment episodes of the LTST DB [1], and automated ischaemic and non-ischaemic heartbeat classification [13]. The Hermite polynomials were used for estimating ECG wave features [14] and for clustering ECG complexes [15]. ...
... The motivation for a new approach using the orthogonal transformation of ST segment based on orthogonal polynomials comes from observing the shapes of the ST segment KLT basis functions [7] obtained from the European Society of Cardiology ST-T Database (ESC DB) [2,22], the standard reference for assessing the quality of AECG analyzers. These basis functions (see Fig 3) span over two ECG leads. ...
... To construct the robust covariance matrix, we used clean heart beats from the LTST DB left after preprocessing the records with robust KLT feature-space based noise and the outlier extraction procedure [7]. The procedure proved to be robust and accurate. ...
Differentiation between ischaemic and non-ischaemic transient ST segment events of long term ambulatory electrocardiograms is a persisting weakness in present ischaemia detection systems. Traditional ST segment level measuring is not a sufficiently precise technique due to the single point of measurement and severe noise which is often present. We developed a robust noise resistant orthogonal-transformation based delineation method, which allows tracing the shape of transient ST segment morphology changes from the entire ST segment in terms of diagnostic and morphologic feature-vector time series, and also allows further analysis. For these purposes, we developed a new Legendre Polynomials based Transformation (LPT) of ST segment. Its basis functions have similar shapes to typical transient changes of ST segment morphology categories during myocardial ischaemia (level, slope and scooping), thus providing direct insight into the types of time domain morphology changes through the LPT feature-vector space. We also generated new Karhunen and Lo ève Transformation (KLT) ST segment basis functions using a robust covariance matrix constructed from the ST segment pattern vectors derived from the Long Term ST Database (LTST DB). As for the delineation of significant transient ischaemic and non-ischaemic ST segment episodes, we present a study on the representation of transient ST segment morphology categories, and an evaluation study on the classification power of the KLT- and LPT-based feature vectors to classify between ischaemic and non-ischaemic ST segment episodes of the LTST DB. Classification accuracy using the KLT and LPT feature vectors was 90% and 82%, respectively, when using the k-Nearest Neighbors (k = 3) classifier and 10-fold cross-validation. New sets of feature-vector time series for both transformations were derived for the records of the LTST DB which is freely available on the PhysioNet website and were contributed to the LTST DB. The KLT and LPT present new possibilities for human-expert diagnostics, and for automated ischaemia detection.
... Early computer-aided programs often use manually extracted morphological f tures to compare the threshold [29,30], or the morphological features extracted manua are classified by the machine learning method [10,[31][32][33][34][35]. The diagnosis process can us ally be divided into four steps [1], followed by pretreatment, waveform detection, featu extraction, and classification, of which the most critical step is feature extraction. ...
... Early computer-aided programs often use manually extracted morphological features to compare the threshold [29,30], or the morphological features extracted manually are classified by the machine learning method [10,[31][32][33][34][35]. The diagnosis process can usually be divided into four steps [1], followed by pretreatment, waveform detection, feature extraction, and classification, of which the most critical step is feature extraction. ...
Heart diseases such as myocardial ischemia (MI) are the main causes of human death. The prediction of MI and arrhythmia is an effective method for the early detection, diagnosis, and treatment of heart disease. For the rapid detection of arrhythmia and myocardial ischemia, the electrocardiogram (ECG) is widely used in clinical diagnosis, and its detection equipment and algorithm are constantly optimized. This paper introduces the current progress of portable ECG monitoring equipment, including the use of polymer material sensors and the use of deep learning algorithms. First, it introduces the latest portable ECG monitoring equipment and the polymer material sensor it uses and then focuses on reviewing the progress of detection algorithms. We mainly introduce the basic structure of existing deep learning methods and enumerate the internationally recognized ECG datasets. This paper outlines the deep learning algorithms used for ECG diagnosis, compares the prediction results of different classifiers, and summarizes two existing problems of ECG detection technology: imbalance of categories and high computational overhead. Finally, we put forward the development direction of using generative adversarial networks (GAN) to improve the quality of the ECG database and lightweight ECG diagnosis algorithm to adapt to portable ECG monitoring equipment.
... Ischemic events (ISE) can be confused in an automatic detector by a number of non-ischemic events (NISE) such as transient ST shifts (transient events, TE) due to heart rate increments or by sudden ST shifts (sudden step events, SSE), due to changes in the intraventricular conduction or to changes in the cardiac electrical axis [12]. Heart rate events (HRE) arise when heart rate increases and, as a consequence, the RR interval is shortened and the T wave moves closer to the QRS complex, producing a distortion in the ST segment level measurements [2]. ...
... The limitation of the ESC DB is mainly due to the relatively short length of the records, not including sufficient numbers of NISE to test adequately the specificity of automated ischemic detectors [13]. Moreover, when successfully used on the ESC DB, the performance of algorithms suffers when applied on the LTST DB, specially the predictivity [12,21]. Few works treated the events classification with the LTST DB [2,18]. ...
Holter recordings are widely used to detect cardiac events that occur transiently, such as ischemic events. Much effort has been made to detect early ischemia, thus preventing myocardial infarction. However, after detection, classification of ischemia has still not been fully solved. The main difficulty relies on the false positives produced because of non-ischemic events, such as changes in the heart rate, the intraventricular conduction or the cardiac electrical axis. In this work, the classification of ischemic and non-ischemic events from the long-term ST database has been improved, using novel spectral parameters based on the continuous wavelet transform (CWT) together with temporal parameters (such as ST level and slope, T wave width and peak, R wave peak, QRS complex width). This was achieved by using a nearest neighbour classifier of six neighbours. Results indicated a sensitivity and specificity of 84.1% and 92.9% between ischemic and non-ischemic events, respectively, resulting a 10% increase of the sensitivity found in the literature. Extracted features based on the CWT applied on the ECG in the frequency band 0.5–4 Hz provided a substantial improvement in classifying ischemic and non-ischemic events, when comparing with the same classifier using only temporal parameters.
In this work it is improved the classification of ischemic and non-ischemic events. The main difficulty of ischemic detectors relies on the false positives produced because of non-ischemic events. After a preprocessing stage, temporal and spectral parameters are extracted from events of the Long Term ST Database. The novel parameters proposed in this work are extracted from the Continuous Wavelet Transform. A nearest Neighbor Classifier is used, obtaining a sensitivity and specificity of 84.1% and 92.9%, respectively.
... Ischemia detectors are commonly applied in two scenarios: Holter monitoring, typically during 24 hours, to assess patients with suspected or known coronary artery disease (CAD); and continuous monitoring (e.g. in intensive care units). Most ischemia detectors are based on evaluation of changes in the ST segment deviation of the electrocardiogram (ECG), which has been traditionally considered as the most sensitive marker to diagnose ischemia in clinical practice (Shook et al 1987, Taddei et al 1995, García et al 1998, Jager et al 1998, Stadler et al 2001, Smrdel and Jager 2004. Because ST segment changes may result from many other causes apart from ischemia, such as variations in the electrical heart axis due to body position changes, heart rate-related events, other electrical conduction changes or ECG artifacts, many ST-based ischemia detectors developed are not robust enough to distinguish between ischemic and non-ischemic episodes of ST segment changes (Mincholé et al 2010), which nowadays still remains a challenge. ...
... Results shown in table 2 indicate that the sensitivity and specificity values were mostly lower for the ST level, which could to a certain extent be expected due to the more gradual changes occurring in the ST segment during the occlusion period. These values, however, need to be taken in the context of this detector's focus on abrupt changes, since other ST-based detectors, designed for the more gradual ST changes (Jager et al 1998), could report better ST-based sensitivity and specificity. ...
In this paper, an ischemia detector is presented based on the analysis of QRS-derived angles. The detector has been developed by modeling ischemic effects on the QRS angles as a gradual change with a certain transition time and assuming a Laplacian additive modeling error contaminating the angle series. Both standard and non-standard leads were used for analysis. Non-standard leads were obtained by applying the PCA technique over specific lead subsets to represent different potential locations of the ischemic zone. The performance of the proposed detector was tested over a population of 79 patients undergoing percutaneous coronary intervention in one of the major coronary arteries (LAD (n = 25), RCA (n = 16) and LCX (n = 38)). The best detection performance, obtained for standard ECG leads, was achieved in the LAD group with values of sensitivity and specificity of [Formula: see text], [Formula: see text], followed by the RCA group with [Formula: see text], Sp = 94.4 and the LCX group with [Formula: see text], [Formula: see text], notably outperforming detection based on the ST series in all cases, with the same detector structure. The timing of the detected ischemic events ranged from 30 s up to 150 s (mean = 66.8 s) following the start of occlusion. We conclude that changes in the QRS angles can be used to detect acute myocardial ischemia.
... The model parameters have independent effects on the model output events, i.e. the amplitude, duration, and incidence time of each event can directly be considered as inputs to the program; • This model is capable of generating transient ST-segment episodes such as depression, elevation, and sloped ascending or descending [12] ; • This model consists of algebraic mathematical equations. Therefore, there would be no need to numerical solution routines. ...
... Due to the existence of real infarctions, depolarization changes (QRS complexes) are always followed by abnormal repolarization (ST-T) [7,12] . Necrosis of a rather large part of the myocardial tissue can lead to reductions in R-wave or Q-wave amplitude of anterior, lateral, or inferior leads. ...
In this study, a mathematical model is developed based on algebraic equations which is capable of generating artificially normal events of electrocardiogram (ECG) signals such as P-wave, QRS complex, and T-wave. This model can also be implemented for the simulation of abnormal phenomena of electrocardio-graphic signals such as ST-segment episodes (i.e. depression, elevation, and sloped ascending or descending) and repolarization abnormalities such as T-Wave Alternans (TWA). Event parameters such as amplitude, du-ration, and incidence time in the conventional ECG leads can be a good reflective of heart electrical activity in specific directions. The presented model can also be used for the simulation of ECG signals on torso plane or limb leads. To meet this end, the amplitude of events in each of the 15-lead ECG waveforms of 80 normal subjects at MIT-BIH Database (www.physionet.org) are derived and recorded. Various statistical analyses such as amplitude mean value, variance and confidence intervals calculations, Anderson-Darling normality test, and Bayesian estimation of events amplitude are then conducted. Heart Rate Variability (HRV) model has also been incorporated to this model with HF/LF and VLF/LF waves power ratios. Eventually, in order to demonstrate the suitable flexibility of the presented model in simulation of ECG signals, fascicular ven-tricular tachycardia (left septal ventricular tachycardia), rate dependent conduction block (Aberration), Left Bundle Branch Block (LBBB), Right Bundle Branch Block (RBBB) and acute Q-wave infarctions of inferior and anterior-lateral walls are finally simulated. The open-source simulation code of above abnormalities will be freely available.
... The performance was tested on European ST-T database which achieves the sensitivity of 83.8% and positive predicitivity of 87.1%. Another technique based on principal component analysis by ANN was used in [6]. It employed ST segment feature extraction and radial basis feedback network (RBFN) was subsequently used for classification of ischemic ECG. ...
... The aothors are thankful to the Shivalik Institute of Engineering and Technlogy, Aliyaspur, Ambala, India for financial assitance. [40] 85 86 Geometric method (Taddei et.al) [41] 84 81 PCA (Jagger et.al) [6] 87 88 Wavelet Transform [9] 92 86 Adaptive Logic Network [38] 72 66 Back propagation Network (Magaveras et.al) [3] 89 ...
In recent years several researchers have put great efforts in biomedical engineering for improving the diagnostic techniques used by the physiologists. A lot of research has been done in biomedical signal processing which includes the signal enhancement, signal compression, artifacts and noise removal like power line interference removal, base line drift removal. For detection of cardiac arrhythmia and ischemia using ECG signal, many emerging techniques and algorithms have been proposed. Ischemia is one of the cardiovascular diseases which are responsible for almost 20% of the deaths around the world. Some of the recently developed algorithms by these researchers have given remarkable results for ischemia. In this paper we review these existing algorithms for detection of ischemia in terms of their performance and capabilities with respect to standard databases available worldwide.
... Therefore, the research studies are focused on having a fast and reliable discovery to the trusted and regular detection of ischaemia events from ECG, because ECG is certainly and normally a recorded signal during the patient check-up as well as the examination process. Inside the electrocardiographic (ECG) signal, ischaemia is indicated as slow moving alterations of the ST segment and/or the actual T wave [8][9][10][11][12], as shown in Figure 1. ...
... However, some of the existing methods use neural networks in ischaemia detection. A combination between neural networks with another technique like principal component analysis, various transforms and fuzzy logic improves the sensitivity of the detection [9,21]. Cardiac imaging is also utilized for ischaemia detection by analysing the changes in 3D image of the heart using mathematical equations [22,23]. ...
Abstract This paper highlights a new detection method based on higher spectral analysis techniques to distinguish the Electrocardiogram (ECG) of normal healthy subjects from that with a cardiac ischaemia (CI) patient. Higher spectral analysis techniques provide in-depth information other than available conventional spectral analysis techniques usually used with ECG analysis. They provide information within frequency parts and information regarding phase associations. Bispectral analysis- Bispectrum and Quadratic Phase Coupling techniques are utilized to detect as well as to characterize phase combined harmonics in ECG. The work is developed, tested and validated using Normal Sinus Rhythm Data from the MIT-BIH Database and CI data from the ST Petersburg European ST-T Database. The results validate the efficacy of the introduced method by maintaining 100% sensitivity and achieving 93.33% positive predictive accuracy. The simplicity and robustness of the proposed method makes it feasible to be used within available ECG systems.
... The performance was tested on European ST-T database which achieves the sensitivity of 83.8% and positive predicitivity of 87.1%. Another technique based on principal component analysis by ANN was used in [6]. It employed ST segment feature extraction and radial basis feedback network (RBFN) was subsequently used for classification of ischemic ECG. ...
... The aothors are thankful to the Shivalik Institute of Engineering and Technlogy, Aliyaspur, Ambala, India for financial assitance. [40] 85 86 Geometric method (Taddei et.al) [41] 84 81 PCA (Jagger et.al) [6] 87 88 Wavelet Transform [9] 92 86 Adaptive Logic Network [38] 72 66 Back propagation Network (Magaveras et.al) [3] 89 ...
In recent years several researchers have put great efforts in biomedical engineering for improving the diagnostic techniques used by the physiologists. A lot of research has been done in biomedical signal processing which includes the signal enhancement, signal compression, artifacts and noise removal like power line interference removal, base line drift removal. For detection of cardiac arrhythmia and ischemia using ECG signal, many emerging techniques and algorithms have been proposed. Ischemia is one of the cardiovascular diseases which are responsible for almost 20% of the deaths around the world. Some of the recently developed algorithms by these researchers have given remarkable results for ischemia. In this paper we review these existing algorithms for detection of ischemia in terms of their performance and capabilities with respect to standard databases available worldwide.
... The motion artifacts are dependent on several factors like the type of BMA, the duration and magnitude of BMA and the pace at which these BMAs are performed. Researchers have proposed various methods to detect the motion artifacts and provide a clean ECG signal for the diagnosis12345678910. There are several issues associated while analysing such an ECG signal contaminated with motion artifacts, e.g. ...
... not very satisfactory as the wavelet-based representation does not separate the in-band BMA signal from the ECG. In other works related to BMA analysis from nonambulatory ECG, body position changes are detected for ischemia monitoring8910. Li et al. [11] have proposed a physical activity (PA) recognition algorithm for a wearable wireless sensor network using both ambulatory electrocardiogram (ECG) and accelerometer signals using support vector machine (SVM) and Gaussian mixture models (GMM). Ambulatory ECG signals/recordings have recently been used for several other purposes like classification of paroxysmal and persistent atrial fibrillation [12], automated recognition of obstructive sleep apnea syndrome [13], an embedded mobile ECG reasoning system for elderly patients [14], ECG signal compression and classification [15], heart rate and accelerometer data fusion for activity assessment [16], a patient adaptive profile scheme for ECG beat classification [17], automatic detection of respiratory rate [18], an intelligent telecardiology system to detect atrial fibrillation [19], etc. ...
In this paper, a recursive principal component analysis (RPCA)-based algorithm is applied for detecting and quantifying the motion artifact episodes encountered in an ECG signal. The motion artifact signal is synthesized by low-pass filtering a random noise signal with different spectral ranges of LPF (low pass filter): 0-5 Hz, 0-10 Hz, 0-15 Hz and 0-20 Hz. Further, the analysis of the algorithm is carried out for different values of SNR levels and forgetting factors (α) of an RPCA algorithm. The algorithm derives an error signal, wherever a motion artifact episode (noise) is present in the entire ECG signal with 100% accuracy. The RPCA error magnitude is almost zero for the clean signal portion and considerably high wherever the motion artifacts (noisy episodes) are encountered in the ECG signals. Further, the general trend of the algorithm is to produce a smaller magnitude of error for higher SNR (i.e. low level of noise) and vice versa. The quantification of the RPCA algorithm has been made by applying it over 25 ECG data-sets of different morphologies and genres with three different values of SNRs for each forgetting factor and for each of four spectral ranges.
... Many studies have recognized the opportunity and leveraged ML for classifying MI. Jager and colleagues implemented an unsupervised machine learning approach, the Karhunen-Loève (KL) transform, for the automatic detection of transient ST-segment episodes in continuous ambulatory ECG recordings, and differentiated them from non-ischemic ST events 11,12 . A similar approach was later adopted for classifying subtypes of ischemic heart disease 13 16 . ...
Objective:
Prompt identification and recognization of myocardial ischemia/infarction (MI) is the most important goal in the management of acute coronary syndrome (ACS). The 12-lead electrocardiogram (ECG) is widely used as the initial screening tool for patients with chest pain but its diagnostic accuracy remains limited. There is early evidence that machine learning (ML) algorithms applied to ECG waveforms can improve performance. Most studies are designed to classify MI from healthy controls and thus are limited due to the lack of consideration of ECG abnormalities from other cardiac conditions, leading to false positives. Moreover, clinical information beyond ECG has not yet been well leveraged in existing ML models.
Approach:
The present study considered downstream clinical implementation scenarios in the initial model design by dichotomizing study recordings from a public large-scale ECG dataset into a MI class and a non-MI class with the inclusion of MI-confounding conditions. Two experiments were conducted to systematically investigate the impact of two important factors entrained in the modeling process, including the duration of ECG, and the value of multimodal information for model training. A novel multimodal deep learning architecture was proposed to learn joint features from both ECG and patient demographics.
Main results:
The multimodal model achieved better performance than the ECG-only model, with a mean area under the receiver operating characteristic curve (AUROC) of 92.1% and a mean accuracy of 87.4%, which is on par with existing studies despite the increased task difficulty due to the new class definition. By investigation of model explainability, it revealed the contribution of patient information in model performance and clinical concordance of the model's attention with existing clinical insights.
Significance:
The findings in this study help guide the development of ML solutions for prompt MI detection and move the models one step closer to real-world clinical applications.
... Such a medical examination will increase the burden of medical doctors because of rechecking the ECG data; it is required to improve the accuracy of the automatic detection. In special, it is a crucial task to quickly and automatically find the ST changes associated with heart diseases [5]. We therefore focused on the analytical method to automatically find the ST segment abnormalities as well as the typical heart diseases such as ventricular fibrillation (VF) and abnormal T waves. ...
... (Minchole et al., 2005) and (Garcia et al., 2000), for instance, utilized a 25Hz cutoff frequency linear phase finite impulse response (FIR) filter. In addition, (Jager et al., 1992(Jager et al., , 1998) applied a 6th-order Butterworth low-pass filter with a 55Hz cutoff frequency, while (Safdarian et al., 2014) and (Kora and Kalva, 2015) applied Butterworth and Sgolay filters without specifying the filter frequency and order, respectively. Applying moving average filters, (Safdarian et al., 2014), low-pass filtering can also be done. ...
Interest in research involving health-medical information analysis based on artificial intelligence has recently been increasing. Most of the research in this field has been focused on searching for new knowledge for predicting and diagnosing disease by revealing the relation between disease and various information features of data. However, still needed are more research and interest in applying the latest advanced artificial intelligence-based data analysis techniques to bio-signal data, which are continuous physiological records, such as EEG (electroencephalography) and ECG (electrocardiogram). This study presents a survey of ECG classification into arrhythmia types. Early and accurate detection of arrhythmia types is important in detecting heart diseases and choosing appropriate treatment for a patient.
... Selected features are used in majority of classification techniques as normal or MI/ischemia ECG signal. Thresholding method was widely used for ECG classification in [20,22,45,28,29]. Using K value as 5, KNN classifier was used in [18] to classify the signal using description found in [46]. ...
Cardio-vascular diseases are one of the foremost causes of mortality in today’s world. The prognosis for cardio-vascular diseases is usually done by ECG signal, which is a simple 12-lead Electrocardiogram (ECG) that gives complete information about the function of the heart including the amplitude and time interval of P-QRS-T-U segment. This article recommends a novel approach to identify the location of thrombus in culprit artery using the Information Fuzzy Network (IFN). Information Fuzzy Network, being a supervised machine learning technique, takes known evidences based on rules to create a predicted classification model with thrombus location obtained from the vast input ECG data. These rules are well-defined procedures for selecting hypothesis that best fits a set of observations. Simulation results illustrates that the recommended approach yields an accurateness of 92.30%. This novel approach is shown to be a viable ECG analysis approach for identifying the culprit artery and thus localizing the thrombus.
... In 1995, Taddie A. and colleague proposed 2-lead ECG analysis in time series to measure the episode of ST segment vector [1]. Several research groups applied Karhunen-Loe`ve transform-based algorithm for detection of transient ST segment episodes [2][3][4]. In 2006 Milosavljevic N. and colleague proposed wavelet transform-based technique to extract some characteristic features of ECG to detect ST segment [5]. ...
Currently, cardiac arrhythmia is a major cause of life threatening. Electrocardiogram (ECG) is the most useful physiological signal that is used in clinical diagnosis. Some abnormalities of heart functions can be investigated from ECG morphology. Many research works present that the changing of ST-T complex is a crucial parameter related to myocardial ischemia. Therefore, this paper reports our progress in ST-episode detection using time domain analysis. The database used in this study is European ST-T database from Physionet. As the results, the performance of our proposed technique can correctly detect ST-episode with 91.37% of sensitivity.
... In other works related to BMA analysis from nonambulatory ECG, body position changes are detected for ischemia monitoring in [8]- [10]. [11] have proposed a physical activity (PA) recognition algorithm for a wearable wireless sensor network using both ambulatory electrocardiogram (ECG) and accelerometer signals using support vector machine (SVM) and Gaussian mixture models (GMM). ...
The use of wearable ECG recorders is becoming common nowadays for the people suffering from cardiac disorders. Although it is a convenient option for hospitalization, it has an inherent drawback of recorded ECG being contaminated by motion artifacts due to various body movement activities of the wearer. In this paper, the spectral characteristics of motion artifacts occurring in wearable ECG (W-ECG) signals have been studied using principal component analysis (PCA) and wavelet transform. The residuals of PCA and wavelet transform characterize the spectral behaviour of the motion artifacts occurring in W-ECG signals. The ECG signals have been acquired from Biopac MP-36 system and a self-developed wearable ECG recorder. The performance is evaluated by power spectral density (PSD) plots of PCA residual errors as well as statistical parameters like mean, median and variance of PCA and wavelet residuals. The PSD plots indicate that the peak frequency of the motion artifacts occurring due to various body movements (like left arm up-down, right arm up-down, left and right legs up-down, waist twist, walking and sitting up-down) is located around 5-15 Hz, coinciding with the ECG spectrum.
... Assim, o primeiro passo do nosso algoritmo consiste em fazer uma análise do nível de ruído de sinal, cujo objetivo é classificar o ruído encontrado em cada batimento cardíaco do registro de ECG. O ruído é classificado em quatro grupos (JAGER et al., 1998) ...
Resumo: Este trabalho apresenta um método promissor para a filtragem do sinal ECG, chamado Decomposição de Modo Empírico (EMD, do inglês Empirical Mode Decomposition). O EMD pode ser utilizado para a remoção de ruído de baixa ou de alta frequência em sinais de ECG. A fim de melhorar o desempenho na remoção de ruído de alta frequência, este trabalho propõe um algoritmo original baseado no EMD. O desempenho do método é avaliado em sinais de ECG ruidosos reais selecionados a partir do banco de dados MIT-BIH de arritmia. Os resultados obtidos com o método EMD são comparáveis aos que utilizam métodos baseados em Transformada Wavelet. Palavras-chave: Cancelamento de ruído; Eletrocardiograma; Decomposição de Modo Empírico. USING EMPIRICAL MODE DECOMPOSITION FOR ELECTROCARDIOGRAM DENOISING Abstract: This work presents a promising method for ECG signal denoising called Empirical Mode Decomposition (EMD). The EMD can be applied for removing either low or high frequency noise in ECG signals. In order to improve performance on removing high frequency noise, this work proposes an original algorithm based on the EMD method. The performance of the method is evaluated on real noisy ECG signals selected from the MITH-BIH Arrhythmia Database. The results obtained with the EMD method are comparable to the ones using a Wavelet Transform based denoising method.
... Different studies propose detectors of ischaemic episodes (e.g. [34,65,67]). However, the results which have been obtained until today are not satisfactory and therefore an accurate detection of ischaemic events considering the long-term ECG still constitutes a challenging task [21]. ...
... ECG features can be extracted in time domain, in frequency domain, or represented as statistical measures. Lots of schemes using different techniques have been proposed for feature extraction, such as wavelet transform (WT) [1][2][3][4][5][6][7][8][9][10], principal component analysis (PCA) [11][12][13], KL transforms [14], Hermite function [15], and Hilbert transform [16]. For the design of ECG classification systems, many schemes have also been presented, including morphology method [17], fuzzy inference engines [18], particle swarm optimization [1,19], support vector machines [20][21][22], hidden Markov model [23], independent component analysis [24,25], nearest neighbor method [9], linear discriminant analysis [26,27], and artificial neural network (ANN). ...
A method for electrocardiogram (ECG) pattern modeling and recognition via deterministic learning theory is presented in this paper. Instead of recognizing ECG signals beat-to-beat, each ECG signal which contains a number of heartbeats is recognized. The method is based entirely on the temporal features (i.e., the dynamics) of ECG patterns, which contains complete information of ECG patterns. A dynamical model is employed to demonstrate the method, which is capable of generating synthetic ECG signals. Based on the dynamical model, the method is shown in the following two phases: the identification (training) phase and the recognition (test) phase. In the identification phase, the dynamics of ECG patterns is accurately modeled and expressed as constant RBF neural weights through the deterministic learning. In the recognition phase, the modeling results are used for ECG pattern recognition. The main feature of the proposed method is that the dynamics of ECG patterns is accurately modeled and is used for ECG pattern recognition. Experimental studies using the Physikalisch-Technische Bundesanstalt (PTB) database are included to demonstrate the effectiveness of the approach.
... This process will eliminate noise because KLT is theoretically optimal in terms of separating signal from noise. It is known that five KL coefficients represent the optimal morphology of both of the QRS complex and ST segment in the ECG [3][4]20]. We assumed that it would also separate the whole true ECG signal from noise. ...
... The estimated parameters of the ECG allow us to detect MI [15]. For instance, a Q-wave that is more than 1/4 the size of the S-wave could be an indication of a possible MI [5]. ...
Continuous electrocardiographic (ECG) monitoring using conducting polymer composite sensors (CPS) presents a non-invasive way to detect cardiac irregularities such as myocardial infarction (MI). Electromyography (EMG), which measures muscle activity in the human body, has a frequency range that overlaps that of the ECG wave. As a result, both EMG and ECG data are present when CPSs collect ECG signals. When measuring ECG waves of an individual during motion, we account for EMG by removing the motion artifact from the ECG signal. With the use of a normalized least mean square (NLMS) algorithm and known signal characteristics, we show that EMG noise can be successfully filtered from an ECG signal that is collected using our CPSs in the standard 12 lead ECG placement. Our software produces a diagnostic-friendly ECG signal and then determines the patient's heart rate. When applied to the arrhythmia database from the Massachusetts Institute of Technology and Beth Israel Hospital (MIT-BIH), our heartbeat detection logic has an accuracy of 99.6% with only 199 false beats and 240 missed beats out of 109,494 total heartbeats taken from 48 individual recordings. KEY WORDS Signal processing of physiological signals, wearable devices, biomedical signal processing, medical signal processing.
... able to detect ST depression, they are not as accurate at focusing on ST elevation. [31] 79 HMM [32] 86 KL & Mahalanobis [33] 87 ANN Back propagation [34] Rule-based [17] 89 70 ...
... The plausibility of detecting body position changes from ECG signal has been studied in earlier works related to ischemia monitoring [13], [14], [15], [16], [17], [18]. The common principle of these works is that there is a significant angle shift of heart-axis with change in three different sleeping positions: supine, left lateral and right lateral, which is manifested as an abrupt change in morphology of QRS-complex, S-T segment and T wave. ...
... However, the reported performance is not very satisfactory as the wavelet based representation does not separate the in-band BMA signal from the ECG. In other works related to BMA analysis from non-ambulatory ECG, body position changes are detected for ischemia monitoring in [8][10]. Ming Li et al. [11] have proposed a physical activity (PA) recognition algorithm for a wearable wireless sensor network using both ambulatory electrocardiogram (ECG) and accelerometer signals. ...
... Different ECG changes related to the evolution of ischemia have been described, including T-wave amplitude changes, ST deviations and even alterations in the terminal portion of the QRS complex [13]. Using global representations for the ST-T complex instead of a single point from the ST segment could better characterizes ischemia patterns and yield better identification of occluded artery [14,15]. The most important ECG change associated with ischemia is the ST segment elevation or depression, with depression being most common. ...
In this paper, we propose an algorithm for detection of myocardial ischemic episodes from electrocardiogram (ECG) signal using the wavelet transform technique. The algorithm was tested on data from the European ST-T change database. Results show that this algorithm is effective for distinguishing normal ECGs from ischemic. We developed a method that uses wavelets for extracting ECG patterns that are characteristic for myocardial ischemia.
... shows the sensitivity and specificity of all methods used in this paper. Comparing the results of this table with table 3 which shows the results of other methods introduced in other studies,23,[40][41][42][43] we can conclude that our methods outperformed previous techniques. ...
Ischemic heart disease is one of the common fatal diseases in advanced countries. Because signal perturbation in healthy people is less than signal perturbation in patients, entropy measure can be used as an appropriate feature for ischemia detection.
Four entropy-based methods comprising of using electrocardiogram (ECG) signal directly, wavelet sub-bands of ECG signals, extracted ST segments and reconstructed signal from time-frequency feature of ST segments in wavelet domain were investigated to distinguish between ECG signal of healthy individuals and patients. We used exercise treadmill test as a gold standard, with a sample of 40 patients who had ischemic signs based on initial diagnosis of medical practitioner.
The suggested technique in wavelet domain resulted in the highest discrepancy between healthy individuals and patients in comparison to other methods. Specificity and sensitivity of this method were 95% and 94% respectively.
The method based on wavelet sub-bands outperformed the others.
In this era of stress-filled life styles and cut-throat competitions, cardiovascular diseases and heart abnormalities are becoming common in the people of early age groups. In order to detect the cardiac abnormalities, if any, earlier and get properly treated it is necessary to have routine body check-ups and even hospitalization at regular intervals. However, due to time and other constraints it may not be possible, for everyone who is at potential cardiac risk, to maintain regularity in this respect. An easy and convenient option to hospitalization is to use the wearable devices (WD). A compact, light-weighted, rugged and full of features yet affordable WDs are now available for healthcare monitoring. These devices are capable of monitoring and recording vital physiological parameters like body temperature, blood pressure, heart rate and many more for hours and even for days. One such physiologically useful and important device is wearable ambulatory electrocardiogram recorder popularly known as W-ECG or A-ECG recorder. The modern W-ECG/A-ECG recorders not only record the ECG signals and related physiological parameters, but are also capable, due to advancements in telemedicine, of updating the physician whenever an abnormal cardiac event or arrhythmia occurs to the wearer.
In this study we have focused on such a wearable ambulatory ECG (A-ECG) recorder and its implications on the recorded A-ECG signals. Although the A-ECG recorders have numerous advantages, the most important being a convenient option to hospitalization, it has several drawbacks as well. The most prominent drawback of an A-ECG recorder is the motion artifacts induced in the recorded A-ECG signals due to various physical activities (PAs) or body movement activities (BMAs) of the subject or the wearer like arms movements, legs movements, walking, climbing stairs up/down, twisting waist or neck, sitting down/standing up or even some light exercises like cycling, jumping, stretching etc. The prime objective of this study is to investigate the impact analysis of these BMAs on A-ECG signal the classification of various types of BMAs. There are several issues related to the A-ECG signal and the motion artifacts associated with it. We mainly focused on the detection of motion artifacts episodes in the A-ECG signals and its impact analysis; the spectral study of motion artifacts; extraction of motion artifacts from A-ECG signals; extracting peculiar features from motion artifact signal and proposing various methods for BMA classification.
For detection of motion artifacts and its impact analysis we implemented a recursive principal component analysis (RPCA) based algorithm suggested by Pawar et al. [7], [8]. The RPCA algorithm works on the ECG beats and not on the samples of ECG signal; hence, it is necessary to generate aligned ECG beats of fixed size, with respect to a common R-peak, from the input ECG samples. This requires accurate QRS detection; hence first and second derivative based QRS complex detection algorithms have been implemented as a prerequisite of RPCA algorithm realization. The artifact episodes have been synthetically generated by low-pass filtering the Gaussian noise of three different SNR levels (variable) and four bandwidths (0-5 Hz, 0-10 Hz, 0-15 Hz and 0-20 Hz) and mixed with the ECG signals, available from MIT-BIH arrhythmia database on Physionet. The impact analysis and quantification of RPCA algorithm has been carried out by performing 25 × 36 experiments (36 simulations‒ four noise bandwidths, three SNRs and three forgetting factors‒ on 25 ECG signals).
The spectral characteristics of motion artifacts contained in A-ECG signals, recorded by the Biopac MP 36 data acquisition system and the wearable ECG recorder, have been studied using the principal component analysis (PCA) and Wavelet transform based approaches. The A-ECG signals each of duration 300 seconds and in lead II configuration with following BMAs: left arm up-down movement, right arm up-down movement, waist-twist movement and sitting down-standing up walking / movement of five healthy subjects have been recorded using these recorders. The residuals, obtained by applying PCA on recorded A-ECG signals with 5, 10 and 15 principal components, have been regarded as motion artifact signals. Similarly, in wavelet transform based approach the A-ECG signals have been decomposed upto fifth level using ‘bior 3.7’, ‘symlet 4’ and ‘coiflet 5’ wavelets and the motion artifacts have been obtained by collecting the wavelet residuals. The spectral characteristics of the motion artifacts signals obtained by the two approaches have been studied using power spectral density (PSD) plots.
In order to classify the BMAs of an individual person, it is necessary to extract the predominant time/frequency features of motion artifact signals. These features have been extracted using Gabor transform and these feature vectors have been fed to three different types of classifiers: artificial neural network (ANN), neuro-fuzzy classifier (NFC) and support vector machine (SVM) for BMA classification. Using these classifiers single-fold and ten-fold validation experiments have been performed for BMA classification of five subjects. Although, all the three classifiers have achieved an overall classification rate of over 95%, i.e. only 5% of wrong or misclassification, it is the time in which the classification is achieved distinguishes them. It is observed that the NFC due to its more complex structure takes highest classification time, whereas the SVM is the fastest of them all with more rugged and consistent classification performance.
Elevation or depression in an electrocardiographic ST segment is an important indication of cardiac Ischemia. Computer-aided algorithms have been proposed in the recent past for the detection of ST change in ECG signals. Such algorithms are accompanied by difficulty in locating a functional ST segment from the ECG. Laborious signal processing tasks have to be carried out in order to precisely locate the start and end of an ST segment. In this work, we propose to detect ST change from heart rate variability (HRV) or RR-interval signals, rather than the ECG itself. Since HRV analysis does not require ST segment localization, we hypothesize an easier and more accurate automated ST change detection here. We use the recent concept of entropy profiling to detect ST change from RR interval data, where the estimation corresponds to irregularity information contained in the respective signals. We have compared results of SampEn, FuzzyEn and TotalSampEn (entropy profiling) on 18 normal and 28 ST-changed RR interval signals. SampEn and FuzzyEn give maximum AUCs of 0.64 and 0.62 respectively, at the data length N = 750. T otalSampEn shows a maximum AUC of 0.92 at N = 50, clearly proving its effectiveness on short-term signals and an AUC of 0.88 at N = 750, proving its efficiency over SampEn and F uzzyEn.
Cardiovascular diseases are one of the main causes of death in the world, as a result much efforts have been made to detect early ischemia. Traditionally changes produced in the ST or STT segments of the heartbeat were analyzed. The main difficulty relies on alterations produced in the ST or STT segment because of non ischemic events, such as changes in the heart rate, the ventricular conduction or the cardiac electrical axis. The aim of this work is to differentiate between ischemic and heart rate related events using the information provided by the continuous wavelet transform of the electrocardiogram. To evaluate the performance of the classifier, the Long Term ST Database was used, with ischemic and non ischemic differentiated events annotated by specialists. The analysis was performed over 77 events (52 ischemic and 25 heart rate related), obtaining a sensitivity and positive predictivity of 86.64% for both indicators.
myocardial ischemic episode () MI (ﻗﻠﺒﻲ ﮔﺮدﺷﻲ ﻣﺎﻧﻴﺘﻮرﻳﻨﮓ در ECG) AECG (ﺗﻐﻴﻴﺮات ﭘﺎﻳﻪ ﺑﺮ ST-Segment ﻣﻌﺮﻓ اﺳﺖ ﺷﺪه ﻲ. اﺳﺎس ﺑﺮ روش اﻳﻦ ﺗﺒ ﺪ ﮔﺴﺴﺘﻪ ﻛﺴﻴﻨﻮﺳﻲ ﻳﻞ Discrete Cosine Transform) DCT (و ﻋﺼﺒﻲ ﺷﺒﻜﻪ Artificial Neural Network) ANN (اﺳﺖ ﺷﺪه اراﺋﻪ و ارزﻳﺎﺑﻲ. اﺑﺘﺪا در ST-Segment ﺗﺸﺨﻴﺺ اﺳﺎس ﺑﺮ ﻣﻜﺎن ﭘﻴﻚ R در AECG اﺳﺘﺨﺮاج ﻣﻴﺸﻮد. ﺿﺮاﻳﺐ از اي ﻣﺠﻤﻮﻋﻪ زﻳﺮ DCT ﺑﺮدار ﺑﻌﻨﻮان از ﺣﺎﺻﻞ وﻳﮋﮔﻲ ST-Segment اي ﻻﻳﻪ ﺳﻪ ﺷﺒﻜﻪ ﻳﻚ ﻧﻬﺎﻳﺖ در و ﻣﻴﺸﻮد ﺗﻌﻴﻴﻦ Feed Forward ﺑﺎ اﻟﮕﻮرﻳﺘﻢ از ﮔﻴﺮي ﺑﻬﺮه Backpropagation ﺑﻨﺪي ﻛﻼﺳﻪ ﺑﺮاي ST-Segment ﻧﺮﻣﺎل ﻏﻴﺮ ﻳﺎ و ﻧﺮﻣﺎل ﺑﻌﻨﻮان) ﺑﻪ ﻣﺒﺘﻼ ﺑﻴﻤﺎران در MI (ﻣﻴﺸﻮد اﺳﺘﻔﺎده. در ﺑﺎ ﺑﻨﺪي ﻃﺒﻘﻪ ﻧﺮخ ﻛﺎﻣﭙﻴﻮﺗﺮي ﺳﺎزي ﺷﺒﻴﻪ ﺣﺪود در ﻻﻳﻲ 82 % آﻣﺪ ﺑﺪﺳﺖ. ﻛﻪ داد ﻧﺸﺎن آﻣﺪه ﺑﺪﺳﺖ ﻧﺘﺎﻳﺞ DCT ﻗﺎﺑﻞ ﭘﻴﺸﻨﻬﺎد ﻳﻚ ﻋﺼﺒﻲ ﺷﺒﻜﻪ و ﺑﺎ ﺧﻮﻧﻲ ﻛﻢ ﻧﺎرﺳﺎﻳﻲ ﺗﺸﺨﻴﺺ ﺑﺮاي ﻗﺒﻮﻟﻲ ﮔﻴﺮي ﺑﻬﺮه از ST-Segment ﺳﻴﮕﻨﺎل در AECG ﻣﻴﺒﺎﺷﺪ واژه ﻛﻠﻴﺪ-Artificial Neural Network-Backpropagation-Discrete Cosine Transform-Myocardial Ischemic Episode-ST-Segment 1-ﻣﻘﺪﻣﻪ Myocardial Ischemia) MI (ﺟﺮﻳﺎن ﻛﻪ اﻳﺴﺖ ﻋﺎرﺿﻪ ﺑﻪ و ﺧﻮﻧﺮﺳﺎﻧﻲ ﺗ ﺒﻊ آ ﻗﻠﺒﻲ ﻋﻀﻼت ﺑﻪ رﺳﺎﻧﻲ اﻛﺴﻴﮋن ن ﻣﺨﺘﻞ ﻣﻲ ر روي آن ﺗﺎﺛﻴﺮ ﻛﻪ ﺷﻮد و در ﺑﻄﻨﻲ ﭘﻼرﻳﺰاﺳﻴﻮن ﺿﺮﺑﺎن ﻫﺮ ﻣﻲ ﺷﺪن ﻃﺒﻴﻌﻲ ﻏﻴﺮ ﺑﺎﻋﺚ و ﺑﺎﺷﺪ ST-Segment ﺷﻜﻞ در ﻣﻮج ECG ﻣﻲ ﺷﻮد. ﺗﺸﺨﻴﺺ ﻧﺎرﺳﺎﻳﻲ اﻳﻦ اﻫﻤﻴﺖ و ﺷﺪه ﻗﻠﺒﻲ ﻋﻀﻼت ﮔﺮﻓﺘﮕﻲ ﺑﺎﻋﺚ ﻛﻪ ﭼﺮا دارد اي وﻳﮋه اﺳﺖ ﻗﻠﺒﻲ ﺣﻤﻼت ﻋﻮاﻣﻞ از ﻳﻜﻲ. ﺣﻤﻼت ﻃﻮل در MI در ﺗﻐﻴﻴﺮات ST-Segment ﺗﻐﻴﻴﺮات از ﺑﻴﺸﺘﺮ ﻣﺘﻨﺎوب ﺑﻄﻮر ﻋﺎدي ﺣﺎﻟﺖ در ﻣﻲ ﺑﺎﺷﺪ. در ﺗﻐﻴﻴﺮات از ﺑﺮﺧﻲ ﺷﺎﻳﺪ ﺑﻨﺎﺑﺮاﻳﻦ زﻣﺎﻧﻲ ﺑﺎزه ﻛﻮﺗ ﺎه ECG ﻧ ﺗﺸﺨﻴﺺ ﻗﺎﺑﻞ ﺒﺎﺷﺪ. ارزﻳﺎﺑﻲ ﺑﺮاي ﻧﺎرﺳﺎﻳﻲ اﻳﻦ از ﺻﺤﻴﺢ ﻣﻮﻧﻴﺘﻮرﻳﻨﮓ ، ﺑﻴﻤﺎر از ﻣﺪت ﻃﻮﻻﻧﻲ اﻧﺠﺎم را ﺧﻮد روزﻣﺮه ﻓﻌﺎﻟﻴﺘﻬﺎي ﺑﻴﻤﺎر ﺣﺎﻟﻴﻜﻪ در ﻣﻲ دﻫﺪ اﺳﺖ ﻧﻴﺎز ﭘﺰﺷﻜﻲ ﻣﺘﺨﺼﺼﻴﻦ ، ECG ﺗﺤﺖ ﻃﻮﻻﻧﻲ زﻣﺎن ﻃﻮل در را ﻋﻨﻮان ambulatory electrocardiogram) AECG (ﺿﺒﻂ ﻣﻲ ﻛﻨﻨﺪ. ﻃﺒﻘﻪ روﺷﻬﺎي ﭘﻴﺸﺒﺮد در ﺗﻮﺟﻬﻲ ﻗﺎﺑﻞ ﺗﺤﻘﻴﻘﺎت ﺑﻨﺪي ST-Segment ﺻﻮرت ﻧﺮﻣﺎل ﻏﻴﺮ ﻳﺎ و ﻧﺮﻣﺎل ﺑﻌﻨﻮان اﺳﺖ ﮔﺮﻓﺘﻪ [1]. ﺳﺮي ﻳﻚ ﺑﻪ روﺷﻬﺎ اﻳﻦ ﺗﺴﺖ و آﻣﻮزش
Cardiac disease is one of the leading causes of death all over the world. With the inception of fast signal processing and computing hardware, techniques for the automatic detection of cardiac disorders through ECG has stemmed up as one of the most promising methodologies in Clinical Decision Support Systems. Such a system can offer rapid, accurate and reliable diagnosis to a variety of cardiac diseases and can reduce the work load for cardiac experts along with providing a facility for the simultaneous monitoring of multiple patients. In this work we have developed techniques for the automatic processing and analysis of the ECG. The work is divided into three major parts: Part-I involving study and implementation of methods for removal of artifacts from the ECG. These include baseline and noise removal techniques. In this work we have compared different baseline removal techniques, such as use of digital FIR and IIR filters and 3 different polynomial fitting approaches, to find out that the use of a two stage first order polynomial fitting based method introduces least distortion in the ECG while effectively compensating the ECG baseline. For Noise removal, we compare and contrast three different techniques, i.e. Use of Digital filters, Independent Component Analysis (ICA) and Local Nonlinear Projective Filtering. We conclude that nonlinear projective filtering performs well in removing noise from the ECG, whereas the potential of ICA for this purpose has been explored.
Part-II involves the segmentation of different ECG components, i.e. P, QRS and Twaves using methods based on digital filters, Continuous Wavelet Transform (CWT) and the Discrete Wavelet Transform. A new method for QRS detection and delineation through CWT has been developed which compares well with existing research offering Sensitivity/Specificity of ~99.8% for detection of QRS with ~10ms error in determining its onset and offset. The accuracy of an existing DWT based method has been improved through the use of Genetic Algorithms (GA). We conclude that the use of DWT with parameter optimization through GA proves to be the most effective technique for ECG Segmentation giving equally good accuracy in terms of detection and delineation.
Part-III is concerned with the classification of different types of heart rhythms (Normal, Atrial Premature Beats, Ventricular Premature Beats, Paced Rhythms, Left xvii and Right Bundle Branch Blocks) and the detection of ST Segment deviations connected to Ischemic Heart Disease. For the purpose of classification of different arrhythmias we have compared DWT based features with those obtained from the Discrete Fourier Transform (DFT) to conclude that DWT is more effective in the classification of different types of heart rhythms. We have achieved 99.1% accuracy through implementing a DWT based technique for feature extraction and using k- Nearest Neighbor classifiers. These results have been compared with those obtained through the use of Probabilistic Neural Networks (PNN) and Learning Vector Quantization (LVQ) Neural Networks. We have also compared the performance of different types of feature extraction and classification techniques for the detection of ischemic ST deviation episodes, such as time-domain features with a rule based classifier, use of Principal Component Analysis (PCA) based features with a Backpropagation Neural Network, a Neural Network Ensemble and a Support Vector Machine (SVM) ensemble classifier. We have achieved a Sensitivity/Positive Predictivity of ~90% with the use of a novel Neural Network Ensemble which uses lead specific principal components as features. These results are highest in terms of accuracy when compared with the existing literature with the novelty lying in the use of lead specific KLT Bases and Ensemble Neural Classifiers for each lead.
The work reported in this thesis can be used to establish the foundations of a practical stand-alone system for patient monitoring and the design of a multiple patient monitoring system as required in hospitals.
Cardiovascular function reflects the overall ability of the two cardiac pumps and their circulation to secure tissular functionality by providing an adjusted supply of oxygen and nutrients while eliminating biological waste products. Failure of this ability causes severe damage and, if protracted and/or extensive, may be lethal. Leading causes of death are cardiovascular (life-threatening dysrhythmia, coronary and cerebrovascular atherosclerotic vasculopathies, heart [pump] failure, etc.) and related morbidity is highly incapacitating with important negative socio-economic implications. It is therefore self-evident that cardiovascular morbidity and mortality and its precursor conditions (atherosclerosis) or risk factors (hypertension) are important target indications for drug development.
This paper highlights a new method for the detection of ischaemic episodes using statistical features derived from ST segment deviations in electrocardiogram (ECG) signal. Firstly, ECG records are pre-processed for the removal of artifacts followed by the delineation process. Then region of interest (ROI) is defined for ST segment and isoelectric reference to compute the ST segment deviation. The mean thresholds for ST segment deviations are used to differentiate the ischaemic beats from normal beats in two stages. The window characterization algorithm is developed for filtration of spurious beats in ischaemic episodes. The ischaemic episode detection is made through the coefficient of variation (COV), kurtosis and form factor. A bell-shaped normal distribution graph is generated for normal and ischaemic ST segments. The results show average sensitivity (Se) 97.71% and positive predictivity (+P) 96.89% for 90 records of the annotated European ST-T database (EDB) after validation. These results are significantly better than those of the available methods reported in the literature. The simplicity and automatic discarding of irrelevant beats makes this method feasible for use in clinical systems.
In the paper, we present a fully automated real-time multi-lead ST-segment monitoring algorithm. For a representative normal beat in each ST measurement interval, the ECG leads with low signal quality are excluded and the remaining leads are used in a multi-lead waveform-length transformation to form a length signal for Q-onset (Q) and J-point (J) determination. From Q, the isoelectric point is determined and used with the J to measure the ST-segment at J or J plus an offset for all available leads. A development set of 158 records and a test set of 60 records with cardiologists' beat-by-beat Q and J annotations were used to develop and evaluate the Q and J detection. The ESC ST-T Database and a 60-patient annotated 12-lead PTCA dataset were used to evaluate the algorithm's ST performance. Detailed statistical results are given in the paper. The test results demonstrate that the described ST-segment monitoring algorithm is effective and reliable.
This chapter presents the viability analysis and the development of heart disease identification embedded system. It offers a time reduction on electrocardiogram - ECG signal processing by reducing the amount of data samples without any significant loss. The goal of the developed system is the analysis of heart signals. The ECG signals are applied into the system that performs an initial filtering, and then uses a Gustafson-Kessel fuzzy clustering algorithm for the signal classification and correlation. The classification indicates common heart diseases such as angina, myocardial infarction and coronary artery diseases. The system uses the European electrocardiogram ST-T Database - EDB as a reference for tests and evaluation. The results prove the system can perform the heart disease detection on a data set reduced from 213 to just 20 samples, thus providing a reduction to just 9.4% of the original set, while maintaining the same effectiveness. This system is validated in a Xilinx Spartan®-3A FPGA. The FPGA implemented a Xilinx Microblaze® Soft-Core Processor running at a 50 MHz clock rate.
In this era of stress-filled life styles and cut-throat competitions, cardiovascular diseases and heart abnormalities are becoming common in the people of early age groups. In order to detect the cardiac abnormalities, if any, earlier and get properly treated it is necessary to have routine body check-ups and even hospitalization at regular intervals. However, due to time and other constraints it may not be possible, for everyone who is at potential cardiac risk, to maintain regularity in this respect. An easy and convenient option to hospitalization is to use the Wearable Devices (WD). A compact, light weighted, rugged and full of features yet affordable WDs are now available for healthcare monitoring. These devices are capable of monitoring and recording vital physiological parameters like body temperature, blood pressure, heart rate and many more for hours and even for days. One such physiologically useful and important device is wearable ambulatory electrocardiogram recorder popularly known as W-ECG or A-ECG recorder. The modern W-ECG/A-ECG recorders not only record the ECG signals and related physiological parameters, but are also capable due to advancements in telemedicine of updating the physician whenever an abnormal cardiac event or arrhythmia occurs to the wearer.
In this study we have focused on such a wearable Ambulatory ECG (A-ECG) recorder and its implications on the recorded A-ECG signals. Although the A-ECG recorders have numerous advantages, the most important being a convenient option to hospitalization, it has several drawbacks as well. The most prominent drawback of an A-ECG recorder is the motion artifacts induced in the recorded A-ECG signals due to various Physical Activities (PAs) or Body Movement Activities (BMAs) of the subject or the wearer like arms movements, legs movements, walking, climbing stairs up/down, twisting waist or neck, sitting down/standing up or even some light exercises like cycling, jumping, stretching etc. The prime objective of this study is to investigate the impact analysis of these BMAs on A-ECG signal the classification of various types of BMAs. There are several issues related to the A-ECG signal and the motion artifacts associated with it. We mainly focused on the detection of motion artifacts episodes in the A-ECG signals and its impact analysis; the spectral study of motion artifacts; extraction of motion artifacts from A-ECG signals; extracting peculiar features from motion artifact signal and proposing various methods for BMA classification.
For detection of motion artifacts and its impact analysis we implemented a Recursive Principal Component Analysis (RPCA) based algorithm suggested by Pawar et al., [1] [2]. The RPCA algorithm works on the ECG beats and not on the samples of ECG signal; hence, it is necessary to generate aligned ECG beats of fixed size, with respect to a common R-peak, from the input ECG samples. This requires accurate QRS detection, hence first and second derivative based QRS complex detection algorithms have been implemented as a prerequisite of RPCA algorithm realization. The artifact episodes have been synthetically generated by low-pass filtering the Gaussian noise of three different SNR levels (variable) and four bandwidths (0-5 Hz, 0-10 Hz, 0-15 Hz and 0-20 Hz) and mixed with the ECG signals, available from MIT-BIH arrhythmia database on Physionet. The impact analysis and quantification of RPCA algorithm has been carried out by performing 25×36 experiments (36 simulations four noise bandwidths, three SNRs and three forgetting factors on 25 ECG signals). The spectral characteristics of motion artifacts contained in A-ECG signals, recorded by the Biopac MP 36 data acquisition system and the wearable ECG recorder, have been studied using the Principal Component Analysis (PCA) and Wavelet transform based approaches. The A-ECG signals each of duration 300 seconds and in lead II configuration with following BMAs: left arm up-down movement, right arm up-down movement, waist-twist movement and sitting down-standing up walking/movement of five healthy subjects have been recorded using these recorders. The residuals, obtained by applying PCA on recorded A-ECG signals with 5, 10 and 15 principal components, have been regarded as motion artifact signals. Similarly, in wavelet transform based approach the A-ECG signals have been decomposed upto fifth level using ‘bior 3.7’, ‘symlet 4’ and ‘coiflet 5’ wavelets and the motion artifacts have been obtained by collecting the wavelet residuals. The spectral characteristics of the motion artifacts signals obtained by the two approaches have been studied using Power Spectral Density (PSD) plots.
In order to classify the BMAs of an individual person, it is necessary to extract the predominant time/frequency features of motion artifact signals. These features have been extracted using Gabor transform and these feature vectors have been fed to three different types of classifiers: Artificial Neural Network (ANN), Neuro-Fuzzy Classifier (NFC) and Support Vector Machine (SVM) for BMA classification. Using these classifiers single-fold and ten-fold validation experiments have been performed for BMA classification of five subjects. Although, all the three classifiers have achieved an overall classification rate of over 95%, i.e. only 5% of wrong or misclassification, it is the time in which the classification is achieved distinguishes them. It is observed that the NFC due to its more complex structure takes highest classification time, whereas the SVM is the fastest of them all with more rugged and consistent classification performance.
Objective: The purpose of this study was to determine the relative importance of bradykinin and nitric oxide (NO) in mediating renal responses to altered sodium intake in Dahl salt-resistant (Dahl-SR) and salt-sensitive (Dahl-SS) rats. Design and methods: Dahl-SR and Dahl-SS rats consumed a diet containing 0.15% (low) or 4.0% (high) sodium chloride for 10 days. A microdialysis technique was then used to measure renal cortical interstitial fluid (RIF) cyclic 39,59-guanosine monophosphate (cGMP) production in anesthetized rats, under baseline conditions and during acute cortical infusion of either the bradykinin B2 receptor antagonist icatibant or the NO synthase inhibitor nitro-L-arginine methyl ester (L-NAME). Urine sodium excretion was monitored simultaneously by ureter cannulation. Results: Baseline sodium excretion was similar in the two types of rats, but RIF cGMP was significantly elevated in Dahl-SR compared to Dahl-SS rats on both low and high sodium diets. Icatibant infusion significantly reduced both RIF cGMP and sodium excretion in Dahl-SR rats during low sodium intake, but had no effect in Dahl-SS rats on either diet. L-NAME infusion significantly reduced sodium excretion in Dahl-SR and Dahl-SS rats, during both low and high sodium intake. L-NAME infusion caused a significant reduction in RIF cGMP in Dahl-SR and Dahl-SS rats on low sodium diet, but reduced RIF cGMP only in Dahl-SR rats on high sodium diet. Conclusion: These data suggest a potential role for cortical bradykinin, but not NO, in mediating the differences in the renal response to low sodium intake between Dahl-SR and Dahl-SS rats.
Myocardial ischemia is a disorder of cardiac function caused by insuficient blood flow to the muscle tissue of the heart. We can diagnose myocardial ischemia by observing the change of ST-segment, but this change is temporary. Our primary purpose is to detect the temporary change of the 57-segment automatically In the signal processing, the wavelet transform decomposes the ECG(electrocardiogram) signal into high and low frequency components using wavelet function. Recomposing the high frequency bands including QRS complex, we can detect QRS complex more easily. Amplitude comparison method is adopted to detect QRS complex. Reducing the effect of noise to the minimum, we grouped ECG by 5 data and compared the amplitude of maximum value. To recognize the ECG .signal pattern, we adopted the polynomial approximation partially and statistical method. The polynomial approximation makes possible to compare some ECG signal with different frequency and sampling period. The ECG signal is divided into small parts based on QRS complex, and then, each part is approximated to the polynomials. After removing the distorted ECG by calculating the difference between the orignal ECG and the approximated ECG for polynomial, we compared the approximated ECG pattern with the database, and we detected and classified abnormality of ECG.
The present contribution presents results of wavelet based methods for computing short-term Heart Rate Variability (HRV) in order to better identify respiratory events by means of analyzing only one lead electrocardiographic (ECG) recordings. Besides RR time interval variability, the performance of other approaches was investigated for the assessment of HRV, such as the intervals between the onsets of successive P waves (PP time series), PR intervals, or ST intervals time series. We analyzed their detection capabilities on respiration events, such as apneas, hypopneas, arterial blood O2 desaturation or arousals, which are used in the diagnoses and characterization of obstructive sleep apnea syndrome (OSAS). The proposed approach gives good results without prior baseline wandering elimination.
In this paper, we present a signal processing method capable of detecting angina in electrocardiograms, and its implementation in Field Programmable Gate Array - FPGA. The adopted procedure is based on fuzzy clustering to reduce the amount of data sampling and a comparison with samples from a previously established database. By using the correlation method on the samples, it is possible to establish an initial indication of angina. The reduced number of samples of the clustering process turns the processing simpler and allows its hardware (FPGA) implementation. According to the tests conducted, the method achieves 85% correct diagnoses.
In this paper, we present a signal processing method capable of detecting angina in electrocardiograms, that was implemented in FPGA. The adopted procedure is based on fuzzy clustering to reduce the amount of data sampling and correlation to compare with samples from a previously established database. By using the correlation method on the samples, it is possible to establish an initial indication of angina. The reduced number of samples of the clustering process turns the processing simpler and allows its hardware implementation in FPGA to validate it. According to the tests conducted, the method achieves 85% correct diagnoses.
In this paper, we present a signal processing method capable of detecting angina in electrocardiograms. The adopted procedure is based on fuzzy clustering to reduce the amount of data sampling and a comparison with samples from a previously established database. By using the correlation method on the samples, it is possible to establish an initial indication of angina. The reduced number of samples of the clustering process turns the processing simpler and allows its hardware implementation. According to the tests conducted, the method achieves 85% correct diagnoses.
Today biomedical signals and image based detection are a basic step to
diagnose heart diseases, in particular, coronary artery diseases. The
goal of this work is to provide non-invasive early detection of Coronary
Artery Diseases relying on analyzing images and ECG signals as a
combined approach to extract features, further classify and quantify the
severity of DCAD by using B-splines method. In an aim of creating a
prototype of screening biomedical imaging for coronary arteries to help
cardiologists to decide the kind of treatment needed to reduce or
control the risk of heart attack.
This supplement follows a format similar to that used in the SEARS Technical Documentation. Chapter 1 consists of a brief description of each of the modifications and then describes the alterations and additions to the relevant model components, subroutines and/or data files. Chapter 2 provides a partial model run showing the user- alterable procedures involved in the implementation of each of these modifications, and chapter 3 contains the schematic flowcharts and supplemental listings of the APL computer code altered as a result of these modifications. This report is part of the overall model documentation. 5 figs.
In this paper QRS complex detection algorithms based on the first and second derivatives have been studied and implemented. The threshold values for detecting R-peak candidate points mentioned in previous work have been modified for accuracy point of view. The derivative based QRS detection algorithms have been found not only computationally simple but exceptionally effective also on variety of ECG database that includes highly noisy and arrhythmic ECG signals. This is indicated by an average detection rate of over 98% obtained through the modified threshold values even for the challenging ECG test sets.
In this paper, the spectral characteristics of
motion artifacts occurring in an ambulatory ECG signal
have been studied using principal component analysis
(PCA). The PCA residual errors characterize the spectral
behavior of the motion artifacts occurring in ambulatory
ECG signals. The ECG signals have been acquired from
Biopac MP-36 system and a self-developed wearable
ECG recorder. The performance is evaluated by power
spectral density (PSD) plots of PCA residual errors as
well as statistical parameters like mean, median and
variance of PCA errors. The PSD plots clearly indicate
that the peak frequency of the motion artifacts occurring
due to various body movements (like left and right arms
up-down, left and right legs up-down, waist twist,
walking and sitting up-down) is located around 20-25 Hz
against the ECG peak frequency around 5-10 Hz.
This article presents the viability analysis and the development of heart disease identification embedded system. It offers a time reduction on electrocardiogram - ECG signal processing by reducing the amount of data samples, without any significant loss. The goal of the developed system is the analysis of heart signals. The ECG signals are applied into the system that performs an initial filtering, and then uses a Gustafson-Kessel fuzzy clustering algorithm for the signal classification and correlation. The classification indicated common heart diseases such as angina, myocardial infarction and coronary artery diseases. The system uses the European electrocardiogram ST-T Database (EDB) as a reference for tests and evaluation. The results prove the system can perform the heart disease detection on a data set reduced from 213 to just 20 samples, thus providing a reduction to just 9.4% of the original set, while maintaining the same effectiveness. This system is validated in a Xilinx Spartan(®)-3A FPGA. The field programmable gate array (FPGA) implemented a Xilinx Microblaze(®) Soft-Core Processor running at a 50MHz clock rate.
Electrocardiogram (ECG) is one of the most common signals used in medical practice because of its noninvasive nature and the information it contains. Its analysis can be used to assess the pathophysiological condition of the heart. Several systems for ECG recording and analysis have been developed for more than a century. Early ECG systems included recording and printing of the signal. Modern systems use computer technology to provide automated diagnosis. The latter is a significant research field and many methods and approaches have been proposed for the detection of ischemia, arrhythmia detection and classification, and diagnosis of chronic myocardial diseases. Those methods are based on the processing of the signal to remove noise and artifacts, extraction of certain features related to diseases, and analysis of the features to obtain the final decision. The analysis is usually based on signal processing, fuzzy logic, and artificial neural networks concepts, mixed with knowledge provided by medical experts. Today, those systems are evaluated using standard databases and must be introduced in clinical practice to be further validated.
The authors describe a two-channel algorithm for robust automated
detection of transient ischemic ST changes. The algorithm operates as a
post-processor to the ARISTOTLE arrhythmia detector. An ST segment
deviation detection function is calculated as the magnitude of the ST
segment vector determined from both leads. Using a variety of auxiliary
functions, the algorithm distinguishes between transient ischemic ST
changes and non-ischemic ST deviations caused by position-related
changes in the electrical axis of the heart
The availability of the European ST-T database makes it possible
to perform quantitative, reproducible performance tests of methods for
detecting and measuring transient ischemic ST changes in the
electrocardiogram (ECG). The authors propose an evaluation protocol and
performance measurements for use with the database. They describe
methods for evaluating the accuracy of ST episode detection, measurement
of ischemia duration, and measurements of ST deviation. Bootstrap
estimation is used to derive expected lower bounds on performance
measures and to assess their utility as predictors of performance. These
methods are illustrated by a case study in which an evaluation is
presented of a two-channel algorithm for automated detection of ischemic
ST episodes
The Karhunen-Loeve transform (KLT) has been used as a tool to analyze the repolarization period in the study of ischemic episodes. The dynamic variations in the ST-T complex shape are shown as variations of the kl series associated with the beat series. Here, the authors propose an adaptive system to estimate the kl coefficients of the ST-T complex in order to improve the signal-to-noise ratio (SNR) of the estimation (it is obtained around 10 dB of improvement). A transversal adaptive linear combiner filter using as reference inputs the KL basis functions and as primary input the concatenation of noisy ST-T completes from consecutive beats is used. It is shown how the weights of the filter become, after convergence, estimates of the kl series. The Least Mean Squares (LMS) and Recursive Least Squares (RLS) algorithms are studied and compared in the kl time series estimation. It is presented a specific initialization for the LMS which leads to the same performance than the RLS.
The authors have begun to develop a new annotated long term ambulatory ST-T database. The aim of the database is to be a reference set containing a number of well documented ischemic ST episodes, axis-related non-ischemic ST episodes, episodes of slow ST level drift and mixed episodes to support development and evaluation of detectors capable of accurate differentiation of ischemic and non-ischemic ST events, as well as basic research into mechanisms and dynamics of ischemia. The authors discuss selection criteria, define the events of interest, and describe the annotation procedure.
Any ST-T segment was here represented by using the principal component analysis, or Karhunen-Loeve Transform (KLT). A representative KL basis set was built from a database containing more than 97000 normal and abnormal ST-T segments. So it was possible to concentrate the 90% of the ST-T signal energy in the first KL coefficients. For the evaluation, the ST-T European Database was chosen, because of its large amount of ischemic episodes. The baseline was removed by using a cubic spline and an adaptive filter was applied in order to improve the signal-to-noise ratio in the final KL series, delivering an improvement of about 10 dB. Then a 3-layers feedforward neural network trained with backpropagation, was applied to the KL series to recognize ST-T level changes. Each input pattern consisted of 28 features, representing 7 ST-T segments, each one described by means of its first 4 KL coefficients. 3 output units were designed, one to describe ST depression, one ST elevation, and one to represent artefacts. The use of principal component analysis and of artificial neural networks allowed us to obtain a sensitivity of 77% and a positive predictive accuracy of 86% on the test set.
The ST-T segment of the surface ECG reflects cardiac repolarization, and is quite sensitive to a number of pathological conditions. ST-T changes generally affect the entire waveshape, and are inadequately characterized by single features such as depression of the ST segment. Metrics which represent overall waveshape should provide more sensitive indicators of ST-T wave abnormalities. This study presents a Karhunen-Loeve Transform (KLT) technique for the analysis of the ST-T waveform. This technique recovers maximum information from a minimum number of parameters for a given set of waveforms. The training data yielded a KL basis set which concentrates 90% of the ST-T signal energy in the first four coefficients. The KL technique was used to analyze the ST-T complexes in the ESC ST-T database. KL coefficients were plotted as a function of time, and were effective in detection of transient ischemic episodes. Twenty percent of the records showed bursts of periodic ischemia suggesting local vascular instability.< >
The authors present an algorithm based on the Karhunen-Loeave
transform (KLT) for robust automated detection of ischemic ST segment
episodes and measurement of the duration of ischemia in two-channel
ambulatory electrocardiographic data. The algorithm operates as a
postprocessor to an existing arrhythmia detector. The episode detector
incorporates a single-scan trajectory recognition technique in the KLT
feature space. The algorithm differentiates between true ischemic ST
segment changes and non-ischemic ST deviations caused by axis shifts. In
evaluations using the European Society of Cardiology ST-T database, the
algorithm achieved a gross ST episode sensitivity of 85.2%, with a
positive predictivity of 86.2%. The gross ischemic ST duration
sensitivity was 75.8%, with a positive predictivity of 78.0%
Several signals are simultaneously acquired during ECG and hemodynamic monitoring in ICU. Relevant features are extracted signals and continuously presented on a visual display unit, usually as trend plots. The real time detection of acute patient status changes is often unreliable because of the high number of features displayed. An approach to the problem of easy identification of the episodes is reported. It is based on a dimension reduction of the stochastic process of the feature time series. This detection method would be the first step toward the achievement of a procedure for the automatic alarm generation and for a selective data storage. The method proposed is based on the principal components analysis. The principal component coefficients are computed over the time series formed by the features extracted from the physiological variables selected during an initial basal period. The subsequent evolving time series are continuously transformed using the same coefficients. Data reduction is then obtained taking into account only the first components. This reduction allows the display of the patient status evolution in two dimensions x-y plot.
Principal component (PC) analysis allows the definition of a
transform with optimum coefficients. The PCs of several time series of
the features, extracted from the electrocardiograms (ECGs) or the
arterial pressure, are calculated. With the assumption that the
statistical structure of intra-patient data is going to be stable, the
basis functions of the transforms are captured once, during a basal
interval. A function has been derived from the two first PCs, which
represent an evidence function to be used both for a compact visual
presentation and for the design of an algorithm for automatic episode
detection. The evaluation of the visual presentation has been based on
the annotated signal of the European ST-T database
The authors describe a method that permits the classification and later comparison of ischaemic episodes detected by a system for the automatic monitoring of patients. They want to obtain an evaluation of the seriousness of the episodes as a function of a set of factors, out of which, one of the most important is the existence of previous ischaemic episodes with a similar morphology. They have applied fuzzy techniques in mimicking in the best possible manner the way in which the human expert operates during the evaluation of these episodes.
Describes a system for the automated detection of ischemic changes in double channel ambulatory EGG. The high performance QRS detector, previously designed and tested by the authors, was applied. A subset of ECG records of the European ST-T Database was used for development, whereas the full set of 90 records and selected 24-hour ECG recordings were used for evaluation. Noisy ECG segments were rejected by the QRS detector or processed by the use of median filters. ST segment deviations, measured beat by beat on both channels, were used for detection of ischemic episodes. The time series of parameters were smoothed by nonlinear filters. Episode detection is based on the analysis of the 2-D path of the two ST deviations. Results of performance evaluation on the European ST-T Database are reported.
The accuracy of a computer system that was developed for the analysis of ST segment changes recorded on Holter ambulatory electrocardiographic monitoring tapes was compared with conventional visual scrutiny, beat by beat analog printouts and a commercial J point trend system. The program calculates and plots multiple electrocardiographically derived variables in a high temporal resolution trend format.Fifty tapes of good recording quality obtained from 19 patients (13 with chronic stable angina and 6 with variant angina) were assessed visually and with the computer system; of these, 20 were analyzed by all four techniques. In the 50 tapes, 629 episodes of diagnostic ST segment changes (all true positive results) were identified by using the computer system. In contrast, only 507 were identified by visual scrutiny; none of these 507 episodes was missed by computer analysis. On the 20 tapes assessed using computer, visual, beat by beat analog and J point trend system analysis, 268, 221, 230 and 178 episodes, respectively, were documented. For the four techniques, false negative and positive results were 0, 47, 38 and 90 and 0, 10, 6 and 24, respectively.The results indicate that, of the systems assessed in this study, the computer program provides the highest accuracy for detection of transient ST segment changes. This is probably accreditable to the compact presentation of multiple electrocardiographically derived variables, allowing a detailed quantitative assessment of 24 hour tapes. It is of particular value for pathophysiologic and pharmacologic studies.
The project for the development of the European ST-T annotated Database originated from a 'Concerted Action' on Ambulatory Monitoring, set up by the European Community in 1985. The goal was to prototype an ECG database for assessing the quality of ambulatory ECG monitoring (AECG) systems. After the 'concerted action', the development of the full database was coordinated by the Institute of Clinical Physiology of the National Research Council (CNR) in Pisa and the Thoraxcenter of Erasmus University in Rotterdam. Thirteen research groups from eight countries provided AECG tapes and annotated beat by beat the selected 2-channel records, each 2 h in duration. ST segment (ST) and T-wave (T) changes were identified and their onset, offset and peak beats annotated in addition to QRSs, beat types, rhythm and signal quality changes. In 1989, the European Society of Cardiology sponsored the remainder of the project. Recently the 90 records were completed and stored on CD-ROM. The records include 372 ST and 423 T changes. In cooperation with the Biomedical Engineering Centre of MIT (developers of the MIT-BIH arrhythmia database), the annotation scheme was revised to be consistent with both MIT-BIH and American Heart Association formats.
The value of ambulatory ST-segment monitoring in the detection of underlying coronary artery disease was investigated in one hundred consecutive patients who underwent exercise testing and coronary arteriography for chest pain. Forty-seven also had thallium-201 radioisotope imaging performed
Six of the 26 patients with normal coronary arteries and 52 of the 74 patients with significant coronary artery disease had ST-segment changes during 48 h of ambulatory monitoring (sensitivity 77%). In comparison, the sensitivity of conventional exercise testing was 73% and specificity was 81%. Previous myocardial infarction did not influence the results, but patients with poor left ventricular function more often had absence of ambulatory ST-segment changes. Three-vessel coronary artery disease was detected more efficiently (sensitivity 80%), compared with single vessel disease (sensitivity 50%). Thallium scintigraphy demonstrated defects of uptake in nine patients without ambulatory ST-segment changes (sensitivity 82%, specificity 71%). The majority of these patients had small inferior or posterior defects in thallium uptake
Only one patient with ambulatory ST-segment changes had normal coronary arteries and demonstrable spasm. Thus, ambulatory ST-segment monitoring is as valuable as stress testing in the detection of coronary artery disease and in addition helps detect patients with coronary spasm and normal coronary arteries.
The present work gives an account of basic principles and available techniques for the analysis and design of pattern processing and recognition systems. Areas covered include decision functions, pattern classification by distance functions, pattern classification by likelihood functions, the perceptron and the potential function approaches to trainable pattern classifiers, statistical approach to trainable classifiers, pattern preprocessing and feature selection, and syntactic pattern recognition.
The authors developed a complete two channel ST episode detection system for long term ECG records. To improve the system sensitivity, a high performance QRS detector was implemented and some noise criteria were applied, to reject too noisy measure values (sens: 97.51% PPA: 99.96%). A three layer feedforward Artificial Neural Network (ANN), trained by backpropagation algorithm, was introduced. It processed the inputs (ST amplitude and ST slope, both in absolute value) in a nonlinear way so that the ST episodes became more easily recognizable from ANN output and the system sensitivity resulted improved (sens: 85% PPA; 88% with vs. sens: 78% PPA: 90% without ANN). The training set was built with 3 out of the 50 records of the European Society of Cardiology ST-T Database. The remaining records were used for system evaluation.< >
A supervised neural network (NN) based algorithm was used to
detect ischemic episodes from electrocardiograms (ECGs). The algorithm
is tested on the European ST-T database. The algorithm is very fast in
its recall state due to the NN, and uses the minimum amount of
information, since it is applied on a one-lead ECG. The adaptive
training backpropagation algorithm reduces dramatically the training
time, and makes possible adjustment training. Even though the algorithm
has some problems with detecting the exact onset and end of an ischemic
episode, its performance was encouraging since it had a gross
sensitivity of 84.4% for ischemia episode detection in the 60 out of 90
records on which it was initially tested. Thus, it seems to be suitable
for use in critical care units due to its speed and training
capabilities
The authors describe robust methods for deriving Karhunen-Loeve
(KL) basis functions, which can be used to represent the QRS complex.
Using a five-term KL expansion of a 200-ms interval, which includes the
QRS complex and part of the ST segment, one can represent morphology on
two simultaneous ECG (electrocardiographic) leads with sufficient
fidelity for beat classification. The residual error of the
representation is an ideal estimate of the instantaneous noise content
of the signal and permits identification of events for which the
morphologic information is unreliable. The authors have compared the
performance of the current KL-based arrhythmia analysis program with its
predecessor (which uses a set of time-domain features for morphology
representation but is otherwise identical to the newer program). In
evaluations using the MIT-BIH and AHA (American Heart Association)
databases, and a newly developed database containing approximately 2.5
million annotated beats (including over 80000 premature ventricular
contractions) from 27 long-term ECG recordings, it was found that beat
classification errors using the KL transform were as little as
one-fourth of those for the older program
Prognostic value of angina pectoris during exercise test in patient with myocardial ischemia
- J M R Detry
- A Robert
- R J Luwaert
- J A Melin
- C R Brohet
- M Dekock
- D Hondt
- A M Dewael
- C Vanbutsele
Detry, J. M. R., Robert, A., Luwaert, R. J., Melin, J. A., Brohet, C. R., DeKock, M., D'Hondt, A. M., Dewael, C., and Vanbutsele, R. Prognostic value of angina pectoris during exercise test in patient with myocardial ischemia. In ''Computers in Cardiology 87,'' pp. 321–324. IEEE Press, Washington, DC, 1988.
Analysis of the cardiac repolarization period using the KL transform: Applications on the ST-T database JAGER, MOODY, AND MARK 322 11. Laguna, P. Model-based estimation of cardiovascular features
- P Laguna
- G B Moody
- R G Mark
Laguna, P., Moody, G. B., and Mark, R. G. Analysis of the cardiac repolarization period using the KL transform: Applications on the ST-T database. In ''Computers in Cardiology 94,'' pp. 233–236. IEEE Press, Los Alamitos, CA, 1995. JAGER, MOODY, AND MARK 322 11. Laguna, P. Model-based estimation of cardiovascular features. In ''Analysis of Non-invasive Cardiovascular Signals in Ischemic Heart Disease: Methods, Development and Evaluation,'' pp.
Neural networks in ischemia detection and ECG processing
- N Maglaveras
Maglaveras, N. Neural networks in ischemia detection and ECG processing. In ''Analysis of Non-invasive Cardiovascular Signals in Ischemic Heart Disease: Methods, Development and Evaluation,'' pp. 22. CNR Inst. of Clinical Physiology, Pisa, 1996.
Use of the “bootstrap” to assess the robustness of the performance statistics of the arrhythmia detector
- Albrecht
Albrecht, P., Moody, G. B., and Mark, R. G. Use of the ''bootstrap'' to assess the robustness of the performance statistics of the arrhythmia detector. J. Ambulatory Monitoring 1(2), 171 (1988).
Ambulatory electrocardiographs. ANSI/AAMI EC38- 1994
- American National
- Standard Institute
American National Standard Institute. Ambulatory electrocardiographs. ANSI/AAMI EC38-
1994 American National Standard Institute, Inc., Association for the Advancement of Medical
Instrumentation, 1994.
Model-based estimation of cardiovascular features
- Laguna
Fuzzy-logic approach to ischemia analysis
- Presedo
Prognostic value of angina pectoris during exercise test in patient with myocardial ischemia
- Detry